From a7794e7da94ac8967095dbc79743a7c6b1760160 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 13 May 2024 10:42:01 +0200
Subject: [PATCH 01/30] Started gaf2aln dev
---
gaf2aln.py | 94 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 94 insertions(+)
create mode 100644 gaf2aln.py
diff --git a/gaf2aln.py b/gaf2aln.py
new file mode 100644
index 0000000..fea7d30
--- /dev/null
+++ b/gaf2aln.py
@@ -0,0 +1,94 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+gaf2aln
+Convert gaf alignement to sam or paf
+
+@author: alexis.mergez@inrae.fr
+@version: 0.1
+"""
+
+import numpy as np
+import pandas as pd
+import argparse
+import os
+
+version = "0.1"
+
+## Argument parser
+arg_parser = argparse.ArgumentParser(description='GFAvc: GFA version converter')
+arg_parser.add_argument(
+ "--gfa",
+ "-g",
+ dest = "gfa",
+ required = True,
+ help = "Graph (.gfa v1)"
+ )
+arg_parser.add_argument(
+ "--gaf",
+ "-a",
+ dest = "gaf",
+ required = True,
+ help = "Alignement file (.gaf)"
+ )
+arg_parser.add_argument(
+ "--format",
+ "-f",
+ dest = "format",
+ default = "P",
+ help = "Output file format. (S: sam, P: paf (default))"
+ )
+arg_parser.add_argument(
+ '--version',
+ '-v',
+ action="store_true",
+ dest = "version",
+ help = "Show version"
+)
+args = arg_parser.parse_args()
+
+# Printing version
+if args.version:
+ print(version)
+ os._exit(0)
+
+# Parsing the .gaf file
+with open(args.gaf, 'r') as file:
+ gaf_lines = file.readlines()
+
+gaf_col = [
+ "QRY.NAME", "QRY.LEN", "QRY.START", "QRY.END", "STRAND",
+ "PATH.MATCH", "PATH.LEN", "ALN.START", "ALN.END",
+ "RES.MATCH", "ALN.BLOCK.LEN", "MAPPING.QUAL"
+ ]
+
+# Creating dictionnary to store alignments
+aln_dict = {}
+for line in range(len(gaf_lines)):
+ ## Splitting the line by tabulation
+ line_content = gaf_lines[line][:-1].split('\t')
+
+ ## Adding alignement info to dictionnary
+ aln_dict[f"ALN_{line+1}"] = {
+ gaf_col[i]: line_content[i] for i in range(len(gaf_col))
+ }
+
+ ## Splitting "PATH.MATCH" into a list
+ aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = [
+ int(node_id) for node_id in aln_dict[f"ALN_{line+1}"]["PATH.MATCH"].split(">")[1:]
+ ]
+
+ ## Adding tags
+ aln_dict[f"ALN_{line+1}"]["TAGS"] = ",".join(line_content[13:])
+
+# Getting nodes of interest ids
+
+# Debug
+print(aln_dict)
+
+
+
+# Parsing the .gaf
+with open(args.gfa, 'r') as file:
+ gfa_lines = file.readlines()
+
--
GitLab
From 0658a3f77518ff5b4c49b461a04d910a3aab673d Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Mon, 13 May 2024 12:44:32 +0200
Subject: [PATCH 02/30] Update gaf2aln.py
---
gaf2aln.py | 68 +++++++++++++++++++++++++++++++++++++++++++++++++++---
1 file changed, 65 insertions(+), 3 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index fea7d30..ced7e7f 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -82,13 +82,75 @@ for line in range(len(gaf_lines)):
aln_dict[f"ALN_{line+1}"]["TAGS"] = ",".join(line_content[13:])
# Getting nodes of interest ids
+aln_nodes = np.unique([
+ node_id
+ for aln in aln_dict.keys()
+ for node_id in aln_dict[aln]["PATH.MATCH"]
+])
+print(aln_nodes)
# Debug
print(aln_dict)
-
-
-# Parsing the .gaf
+# Parsing the .gfa
with open(args.gfa, 'r') as file:
gfa_lines = file.readlines()
+# Nodes length dictionnary structured as follow :
+# {<NODE.ID>: <NODE.LENGTH>}
+nodes_length = {}
+# Nodes dictionnary structured as follow :
+# {<ALN.NODE.ID> : {PATHS: {<PATH.NAME>: (start, end)}}}
+nodes = {}
+# Paths dictionnary structured as follow :
+# {<PATH.NAME>: {NODES: {NODE.ID: <NODE.ID>, ORIENT: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
+paths = {}
+# Links dictionnary structured as follow :
+# {<FROM.NODE.ID>: {<TO.NODE.ID>: {FROM.ORIENT: <FROM.ORIENT>, TO.ORIENT: <TO.ORIENT>}}}
+links = {}
+
+for line in gfa_lines:
+ line_content = line[:-1].split("\t")
+ line_id = line_content[0]
+
+ # Segment line
+ if line_id == "S" :
+
+ nodes_length[str(line_content[1])] = len(line_content[2])
+
+ if line_content[1] in aln_nodes:
+ nodes[str(line_content[1])] = {}
+
+ # Link line
+ elif line_id == "L":
+ try :
+ links[str(line_content[1])][str(line_content[3])] = {
+ "FROM": str(line_content[2]),
+ "TO": str(line_content[4])
+ }
+
+ except :
+ links[str(line_content[1])] = {
+ [str(line_content[3])] : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
+ }
+
+ # Path line
+ elif line_id == "P":
+ paths[str(line_content[1])] = {
+ "NODES": [
+ {"NODE.ID": str(node_id[:-1]), "ORIENT": str(node_id[-1])}
+ for node_id in line_content[2].split(',')
+ ],
+ "CIGAR": line_content[3]
+ }
+
+# Getting the start and end position of alignment nodes on each paths
+def getPathPos(node_id, nodes=nodes, nodes_length=nodes_length, links=links, paths=paths)
+
+for node_id in nodes.keys():
+ for path_id in paths.keys():
+ if
+
+
+
+
--
GitLab
From 81bae737dd5a53d0bf976f78dcd532cb2db1a439 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Tue, 14 May 2024 10:19:15 +0200
Subject: [PATCH 03/30] Update gaf2aln.py
---
gaf2aln.py | 141 ++++++++++++++++++++++++++++++++++++++++-------------
1 file changed, 107 insertions(+), 34 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index ced7e7f..823a51d 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -11,6 +11,7 @@ Convert gaf alignement to sam or paf
import numpy as np
import pandas as pd
import argparse
+import concurrent.futures
import os
version = "0.1"
@@ -38,6 +39,15 @@ arg_parser.add_argument(
default = "P",
help = "Output file format. (S: sam, P: paf (default))"
)
+arg_parser.add_argument(
+ "--threads",
+ "-t",
+ dest = "threads",
+ required = False,
+ default = 1,
+ type = int,
+ help = "Number of threads"
+ )
arg_parser.add_argument(
'--version',
'-v',
@@ -103,53 +113,116 @@ nodes_length = {}
# {<ALN.NODE.ID> : {PATHS: {<PATH.NAME>: (start, end)}}}
nodes = {}
# Paths dictionnary structured as follow :
-# {<PATH.NAME>: {NODES: {NODE.ID: <NODE.ID>, ORIENT: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
+# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
paths = {}
# Links dictionnary structured as follow :
# {<FROM.NODE.ID>: {<TO.NODE.ID>: {FROM.ORIENT: <FROM.ORIENT>, TO.ORIENT: <TO.ORIENT>}}}
links = {}
-for line in gfa_lines:
- line_content = line[:-1].split("\t")
- line_id = line_content[0]
-
- # Segment line
- if line_id == "S" :
+# Parsing the gfa
+## Multithreading function
+def parse_gfa_line(gfa_lines, aln_nodes=aln_nodes):
+ _nodes, _nodes_length, _links, _paths = {}, {}, {}, {}
+ for line in gfa_lines:
+ line_content = line[:-1].split("\t")
+ line_id = line_content[0]
- nodes_length[str(line_content[1])] = len(line_content[2])
+ # Segment line
+ if line_id == "S" :
+
+ _nodes_length[str(line_content[1])] = len(line_content[2])
- if line_content[1] in aln_nodes:
- nodes[str(line_content[1])] = {}
-
- # Link line
- elif line_id == "L":
- try :
- links[str(line_content[1])][str(line_content[3])] = {
- "FROM": str(line_content[2]),
- "TO": str(line_content[4])
+ if line_content[1] in aln_nodes:
+ _nodes[str(line_content[1])] = {"PATHS": {}}
+
+ # Link line
+ elif line_id == "L":
+ try :
+ _links[str(line_content[1])][str(line_content[3])] = {
+ "FROM": str(line_content[2]),
+ "TO": str(line_content[4])
+ }
+
+ except :
+ _links[str(line_content[1])] = {
+ str(line_content[3]) : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
+ }
+
+ # Path line
+ elif line_id == "P":
+ _paths[str(line_content[1])] = {
+ "NODES": {
+ str(node_id[:-1]): str(node_id[-1])
+ for node_id in line_content[2].split(',')
+ },
+ "CIGAR": line_content[3]
}
- except :
- links[str(line_content[1])] = {
- [str(line_content[3])] : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
- }
+ return [_nodes, _nodes_length, _links, _paths]
+
+## Parsing subsets
+executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
+res = {}
+quantiles = np.array(range(args.threads+1))/args.threads
+splits = np.quantile(range(len(gfa_lines)), quantiles, method="higher").tolist()
+splits[-1] += 1
+for i in range(len(splits)-1):
+ res[i] = executor.submit(parse_gfa_line, gfa_lines[splits[i]:splits[i+1]])
+executor.shutdown(wait=True)
+
+## Aggregating results
+for _res in res.values():
+ _nodes, _nodes_length, _links, _paths = _res.result()
+
+ for key, value in _nodes.items():
+ nodes[key] = value
+
+ for key, value in _nodes_length.items():
+ nodes_length[key] = value
- # Path line
- elif line_id == "P":
- paths[str(line_content[1])] = {
- "NODES": [
- {"NODE.ID": str(node_id[:-1]), "ORIENT": str(node_id[-1])}
- for node_id in line_content[2].split(',')
- ],
- "CIGAR": line_content[3]
- }
+ for key, value in _links.items():
+ if not key in links.keys():
+ links[key] = value
+ else :
+ for key_sub, value_sub in value.items():
+ links[key][key_sub] = value_sub
+
+ for key, value in _paths.items():
+ paths[key] = value
# Getting the start and end position of alignment nodes on each paths
-def getPathPos(node_id, nodes=nodes, nodes_length=nodes_length, links=links, paths=paths)
+def getPathPos(path_id, nodes=nodes, nodes_length=nodes_length, links=links, paths=paths):
+ cur_pos = 0
+ _dict = nodes.copy()
+ for path_node in paths[path_id]["NODES"].keys():
+ try :
+ _dict[path_node]["PATHS"][path_id] = (cur_pos, cur_pos+nodes_length[path_node])
+ cur_pos += nodes_length[path_node]
+ except :
+ cur_pos += nodes_length[path_node]
+ return _dict
+
+# Collecting positions of nodes within paths
+## Searching in each path
+executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
+res = {}
+for path_name in paths.keys():
+ print(f"Running on {path_name}")
+ res[path_name] = executor.submit(getPathPos, path_name)
+executor.shutdown(wait=True)
+
+## Storing results
+for path in res.keys():
+ _dict = res[path].result()
+ for node_id, path_dict in _dict.items():
+ print(node_id, path_dict)
+ for path_name, coordinates in path_dict["PATHS"].items():
+ nodes[node_id]["PATHS"][path_name] = coordinates
+
+print(nodes)
+
+
-for node_id in nodes.keys():
- for path_id in paths.keys():
- if
--
GitLab
From 0f682ffdf8b00ee009ac89e8e2d8057d56eeb785 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Tue, 14 May 2024 14:57:45 +0200
Subject: [PATCH 04/30] Update gaf2aln.py
---
gaf2aln.py | 166 +++++++++++++++++++++++------------------------------
1 file changed, 72 insertions(+), 94 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index 823a51d..a684dc8 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -63,6 +63,7 @@ if args.version:
os._exit(0)
# Parsing the .gaf file
+print(f"[gaf2aln::GAF Parser] Reading {args.gaf} ...")
with open(args.gaf, 'r') as file:
gaf_lines = file.readlines()
@@ -73,6 +74,7 @@ gaf_col = [
]
# Creating dictionnary to store alignments
+print(f"[gaf2aln::GAF Parser] Extracting alignments ...")
aln_dict = {}
for line in range(len(gaf_lines)):
## Splitting the line by tabulation
@@ -85,7 +87,7 @@ for line in range(len(gaf_lines)):
## Splitting "PATH.MATCH" into a list
aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = [
- int(node_id) for node_id in aln_dict[f"ALN_{line+1}"]["PATH.MATCH"].split(">")[1:]
+ str(node_id) for node_id in aln_dict[f"ALN_{line+1}"]["PATH.MATCH"].split(">")[1:]
]
## Adding tags
@@ -93,16 +95,15 @@ for line in range(len(gaf_lines)):
# Getting nodes of interest ids
aln_nodes = np.unique([
- node_id
+ str(node_id)
for aln in aln_dict.keys()
for node_id in aln_dict[aln]["PATH.MATCH"]
-])
-print(aln_nodes)
+]).tolist()
-# Debug
-print(aln_dict)
+del gaf_lines, gaf_col
# Parsing the .gfa
+print(f"[gaf2aln::GFA Parser] Reading {args.gfa} ...")
with open(args.gfa, 'r') as file:
gfa_lines = file.readlines()
@@ -111,7 +112,9 @@ with open(args.gfa, 'r') as file:
nodes_length = {}
# Nodes dictionnary structured as follow :
# {<ALN.NODE.ID> : {PATHS: {<PATH.NAME>: (start, end)}}}
-nodes = {}
+nodes = {
+ node_id: {"PATHS": {}} for node_id in aln_nodes
+}
# Paths dictionnary structured as follow :
# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
paths = {}
@@ -120,106 +123,81 @@ paths = {}
links = {}
# Parsing the gfa
-## Multithreading function
-def parse_gfa_line(gfa_lines, aln_nodes=aln_nodes):
- _nodes, _nodes_length, _links, _paths = {}, {}, {}, {}
- for line in gfa_lines:
- line_content = line[:-1].split("\t")
- line_id = line_content[0]
-
- # Segment line
- if line_id == "S" :
-
- _nodes_length[str(line_content[1])] = len(line_content[2])
-
- if line_content[1] in aln_nodes:
- _nodes[str(line_content[1])] = {"PATHS": {}}
+print(f"[gaf2aln::GFA Parser] Extracting nodes, paths and links ...")
+for line in gfa_lines:
+ line_content = line[:-1].split("\t")
+ line_id = line_content[0]
+
+ # Segment line
+ if line_id == "S" :
- # Link line
- elif line_id == "L":
- try :
- _links[str(line_content[1])][str(line_content[3])] = {
- "FROM": str(line_content[2]),
- "TO": str(line_content[4])
- }
-
- except :
- _links[str(line_content[1])] = {
- str(line_content[3]) : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
- }
-
- # Path line
- elif line_id == "P":
- _paths[str(line_content[1])] = {
- "NODES": {
- str(node_id[:-1]): str(node_id[-1])
- for node_id in line_content[2].split(',')
- },
- "CIGAR": line_content[3]
+ nodes_length[str(line_content[1])] = len(line_content[2])
+
+ # Link line
+ elif line_id == "L":
+ try :
+ links[str(line_content[1])][str(line_content[3])] = {
+ "FROM": str(line_content[2]),
+ "TO": str(line_content[4])
}
- return [_nodes, _nodes_length, _links, _paths]
-
-## Parsing subsets
-executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
-res = {}
-quantiles = np.array(range(args.threads+1))/args.threads
-splits = np.quantile(range(len(gfa_lines)), quantiles, method="higher").tolist()
-splits[-1] += 1
-for i in range(len(splits)-1):
- res[i] = executor.submit(parse_gfa_line, gfa_lines[splits[i]:splits[i+1]])
-executor.shutdown(wait=True)
-
-## Aggregating results
-for _res in res.values():
- _nodes, _nodes_length, _links, _paths = _res.result()
-
- for key, value in _nodes.items():
- nodes[key] = value
-
- for key, value in _nodes_length.items():
- nodes_length[key] = value
+ except :
+ links[str(line_content[1])] = {
+ str(line_content[3]) : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
+ }
- for key, value in _links.items():
- if not key in links.keys():
- links[key] = value
- else :
- for key_sub, value_sub in value.items():
- links[key][key_sub] = value_sub
+ # Path line
+ elif line_id == "P":
+ paths[str(line_content[1])] = {
+ "NODES": {
+ str(node_id[:-1]): str(node_id[-1])
+ for node_id in line_content[2].split(',')
+ },
+ "CIGAR": line_content[3]
+ }
- for key, value in _paths.items():
- paths[key] = value
+del gfa_lines
# Getting the start and end position of alignment nodes on each paths
-def getPathPos(path_id, nodes=nodes, nodes_length=nodes_length, links=links, paths=paths):
+print(f"[gaf2aln::Graph processing] Computing nodes positions ...")
+for path_name in paths.keys():
+ print(f"[gaf2aln::Graph processing] Running on {path_name} ...")
cur_pos = 0
- _dict = nodes.copy()
- for path_node in paths[path_id]["NODES"].keys():
+ for path_node in paths[path_name]["NODES"].keys():
try :
- _dict[path_node]["PATHS"][path_id] = (cur_pos, cur_pos+nodes_length[path_node])
+ nodes[path_node]["PATHS"][path_name] = (cur_pos, cur_pos+nodes_length[path_node])
cur_pos += nodes_length[path_node]
except :
cur_pos += nodes_length[path_node]
- return _dict
-# Collecting positions of nodes within paths
-## Searching in each path
-executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
-res = {}
-for path_name in paths.keys():
- print(f"Running on {path_name}")
- res[path_name] = executor.submit(getPathPos, path_name)
-executor.shutdown(wait=True)
-
-## Storing results
-for path in res.keys():
- _dict = res[path].result()
- for node_id, path_dict in _dict.items():
- print(node_id, path_dict)
- for path_name, coordinates in path_dict["PATHS"].items():
- nodes[node_id]["PATHS"][path_name] = coordinates
-
-print(nodes)
+# Reconstructing alignments for each path
+print(f"[gaf2aln::Alignment processing] Computing alignments ...")
+for aln_name in aln_dict.keys():
+
+ print(f"[gaf2aln::Alignment processing] Looking into alignment {aln_name} ...")
+ for path_name in paths.keys():
+
+ print(f"[gaf2aln::Alignment processing] Running on {path_name} ...")
+ cur_pos, cur_aln = 0, []
+
+ # Traversing alignment path
+ for node_id in aln_dict[aln_name]["PATH.MATCH"]:
+
+ # Checking if node is traversed by the current path
+ if path_name in nodes[node_id]["PATHS"].keys():
+ try :
+ cur_aln[-1] += [node_id]
+ except :
+ cur_aln.append([node_id])
+
+ else :
+ # Checking for emptyness
+ if not len(cur_aln) or not len(cur_aln[-1]):
+ cur_aln.append([])
+ else :
+
+
+ # Ajouter le noeud au segment contigue ou finir le dernier segment le cas échéant
--
GitLab
From 15e32c6bda5f4b8a99a24c778da233c3ba98e12c Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Tue, 14 May 2024 16:30:50 +0200
Subject: [PATCH 05/30] Update gaf2aln.py
---
gaf2aln.py | 74 +++++++++++++++++++++++++++++++++++++++++-------------
1 file changed, 56 insertions(+), 18 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index a684dc8..cd73cd4 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -62,6 +62,15 @@ if args.version:
print(version)
os._exit(0)
+# Toolbox
+def walk2path(walk):
+ """
+ Takes a walk in a single string and returns a list of nodes id with signs (gfa v1 like)
+ """
+ _ = re.findall(r'>\w+|<\w+', walk)
+ # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'
+ return [f'{elem[1:]}{(elem[0] == ">")*"+"+(elem[0] == "<")*"-"}' for elem in _]
+
# Parsing the .gaf file
print(f"[gaf2aln::GAF Parser] Reading {args.gaf} ...")
with open(args.gaf, 'r') as file:
@@ -84,11 +93,12 @@ for line in range(len(gaf_lines)):
aln_dict[f"ALN_{line+1}"] = {
gaf_col[i]: line_content[i] for i in range(len(gaf_col))
}
-
+
## Splitting "PATH.MATCH" into a list
- aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = [
- str(node_id) for node_id in aln_dict[f"ALN_{line+1}"]["PATH.MATCH"].split(">")[1:]
- ]
+ aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = {
+ str(node_id[:-1]): node_id[-1]
+ for node_id in walk2path(aln_dict[f"ALN_{line+1}"]["PATH.MATCH"])
+ }
## Adding tags
aln_dict[f"ALN_{line+1}"]["TAGS"] = ",".join(line_content[13:])
@@ -97,7 +107,7 @@ for line in range(len(gaf_lines)):
aln_nodes = np.unique([
str(node_id)
for aln in aln_dict.keys()
- for node_id in aln_dict[aln]["PATH.MATCH"]
+ for node_id in aln_dict[aln]["PATH.MATCH"].keys()
]).tolist()
del gaf_lines, gaf_col
@@ -170,34 +180,62 @@ for path_name in paths.keys():
except :
cur_pos += nodes_length[path_node]
+final_aln = []
+
# Reconstructing alignments for each path
print(f"[gaf2aln::Alignment processing] Computing alignments ...")
for aln_name in aln_dict.keys():
print(f"[gaf2aln::Alignment processing] Looking into alignment {aln_name} ...")
+ aln_dict[aln_name]["HAP.MATCH"] = {}
+
for path_name in paths.keys():
print(f"[gaf2aln::Alignment processing] Running on {path_name} ...")
- cur_pos, cur_aln = 0, []
-
+ cur_aln = []
+ in_aln = False # Keeping track of if we are in an alignment
+
# Traversing alignment path
- for node_id in aln_dict[aln_name]["PATH.MATCH"]:
+ for node_id, strand_on_aln in aln_dict[aln_name]["PATH.MATCH"].items():
+ strand_on_path = paths[path_name]["NODES"][node_id]
+
+ # Checking strand
+ if strand_on_aln == strand_on_path :
+ strand = "S"
+ else :
+ strand = "I"
# Checking if node is traversed by the current path
if path_name in nodes[node_id]["PATHS"].keys():
- try :
- cur_aln[-1] += [node_id]
- except :
- cur_aln.append([node_id])
+ if not in_aln:
+ cur_aln.append({
+ "IN": True,
+ "NODES": {
+ node_id: strand
+ }
+ })
+ in_aln = True
+
+ else : cur_aln[-1]["NODES"][node_id] = strand
else :
- # Checking for emptyness
- if not len(cur_aln) or not len(cur_aln[-1]):
- cur_aln.append([])
- else :
-
+ if in_aln or not len(cur_aln):
+ cur_aln.append({
+ "IN": False,
+ "NODES": {
+ node_id: strand
+ }
+ })
+ in_aln = False
+
+ else : cur_aln[-1]["NODES"][node_id] = strand
+
+ aln_dict[aln_name]["HAP.MATCH"][path_name] = cur_aln.copy()
+ print(f"{path_name} :", cur_aln)
+
+ # Traversing the HAP.MATCH to get alignment
+ for
- # Ajouter le noeud au segment contigue ou finir le dernier segment le cas échéant
--
GitLab
From 146d897c7818d60499b5611294ecdc0d2bcb80e5 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Fri, 17 May 2024 10:31:52 +0200
Subject: [PATCH 06/30] Update gaf2aln.py
---
gaf2aln.py | 175 ++++++++++++++++++++++++++++++++++++-----------------
1 file changed, 121 insertions(+), 54 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index cd73cd4..afc61e2 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -13,6 +13,7 @@ import pandas as pd
import argparse
import concurrent.futures
import os
+import re
version = "0.1"
@@ -71,6 +72,35 @@ def walk2path(walk):
# Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'
return [f'{elem[1:]}{(elem[0] == ">")*"+"+(elem[0] == "<")*"-"}' for elem in _]
+def cigar2basealn(cigar):
+ """
+ Takes a CIGAR string and convert it into a list of base level alignment.
+ For example : "345=" -> ["=", "=", ..., "="] of length 345.
+ """
+ _ = re.findall(r'\d+\D', cigar)
+ print(_)
+ final_cigar = []
+ for match in _:
+ final_cigar += [match[-1]]*int(match[:-1])
+
+ print(final_cigar)
+ return final_cigar
+
+def basealn2cigar(base_aln_list):
+
+ last_elem = base_aln_list[0]
+ CIGAR = [[1, last_elem]]
+ for elem in base_aln_list[1:]:
+ if elem == last_elem:
+ CIGAR[-1][0] += 1
+
+ else :
+ CIGAR[-1][0] = str(CIGAR[-1][0])
+ CIGAR.append([1, elem])
+ last_elem = elem
+ CIGAR[-1][0] = str(CIGAR[-1][0])
+ return "".join(["".join(block) for block in CIGAR if block[1] != ""])
+
# Parsing the .gaf file
print(f"[gaf2aln::GAF Parser] Reading {args.gaf} ...")
with open(args.gaf, 'r') as file:
@@ -100,8 +130,11 @@ for line in range(len(gaf_lines)):
for node_id in walk2path(aln_dict[f"ALN_{line+1}"]["PATH.MATCH"])
}
+ ## Adding CIGAR
+ aln_dict[f"ALN_{line+1}"]["RAW.CIGAR"] = line_content[-1]
+
## Adding tags
- aln_dict[f"ALN_{line+1}"]["TAGS"] = ",".join(line_content[13:])
+ aln_dict[f"ALN_{line+1}"]["TAGS"] = ",".join(line_content[13:-1])
# Getting nodes of interest ids
aln_nodes = np.unique([
@@ -168,74 +201,108 @@ for line in gfa_lines:
del gfa_lines
-# Getting the start and end position of alignment nodes on each paths
-print(f"[gaf2aln::Graph processing] Computing nodes positions ...")
+# Creating GA (Graph alignment) dictionnary storing given info :
+# {<ALN_ID> :
+# { <PATH.ID> : [
+# { NODE.ID, P.ORIENT, A.ORIENT, P.POS, A.POS, CG }
+# ]
+# }
+# }
+
+print(f"[gaf2aln::Graph alignment processing] Computing nodes positions in each paths...")
+# Adding nodes positions relative to path
for path_name in paths.keys():
print(f"[gaf2aln::Graph processing] Running on {path_name} ...")
cur_pos = 0
for path_node in paths[path_name]["NODES"].keys():
try :
nodes[path_node]["PATHS"][path_name] = (cur_pos, cur_pos+nodes_length[path_node])
- cur_pos += nodes_length[path_node]
+ cur_pos += nodes_length[path_node]+1
except :
- cur_pos += nodes_length[path_node]
-
-final_aln = []
-
-# Reconstructing alignments for each path
-print(f"[gaf2aln::Alignment processing] Computing alignments ...")
-for aln_name in aln_dict.keys():
-
- print(f"[gaf2aln::Alignment processing] Looking into alignment {aln_name} ...")
- aln_dict[aln_name]["HAP.MATCH"] = {}
+ cur_pos += nodes_length[path_node]+1
- for path_name in paths.keys():
-
- print(f"[gaf2aln::Alignment processing] Running on {path_name} ...")
- cur_aln = []
- in_aln = False # Keeping track of if we are in an alignment
+# Calculating CIGAR for each nodes in each aln
+print(f"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...")
+# Iterating over alignments
+for aln in aln_dict.keys():
+
+ print(f"[gaf2aln::CIGAR processing] Running on {aln} ...")
+ # Getting the list of base level alignement (["=", "X", ...] from "1=1X...")
+ raw_cigar = cigar2basealn(aln_dict[aln]["RAW.CIGAR"])
+
+ cur_pos = 0
+ print(len(raw_cigar))
+ CIGAR={}
+ aln_nodes_id = list(aln_dict[aln]["PATH.MATCH"].keys())
+ for node_id in aln_nodes_id:
+
+ # Starting aln node
+ if node_id == aln_nodes_id[0] :
+ _cigar = basealn2cigar(raw_cigar[
+ 0:nodes_length[node_id]-int(aln_dict[aln]["ALN.START"])
+ ])
+ cur_pos += nodes_length[node_id]-int(aln_dict[aln]["ALN.START"])
+ # Last aln node
+ elif node_id == aln_nodes_id[-1]:
+ _cigar = basealn2cigar(raw_cigar[cur_pos:])
+ else :
+ _cigar = basealn2cigar(raw_cigar[cur_pos:cur_pos+nodes_length[node_id]])
+ cur_pos += nodes_length[node_id]
+
+ CIGAR[node_id] = _cigar
- # Traversing alignment path
- for node_id, strand_on_aln in aln_dict[aln_name]["PATH.MATCH"].items():
- strand_on_path = paths[path_name]["NODES"][node_id]
+ aln_dict[aln]["CIGAR"] = CIGAR
+ print(CIGAR)
- # Checking strand
- if strand_on_aln == strand_on_path :
- strand = "S"
- else :
- strand = "I"
-
- # Checking if node is traversed by the current path
- if path_name in nodes[node_id]["PATHS"].keys():
- if not in_aln:
- cur_aln.append({
- "IN": True,
- "NODES": {
- node_id: strand
- }
- })
- in_aln = True
-
- else : cur_aln[-1]["NODES"][node_id] = strand
+GA = {}
+# Computing alignments nodes positions in paths
+print(f"[gaf2aln::Graph alignment processing] Lifting alignments coordinates paths positions...")
+for aln_name in aln_dict.keys():
+ GA[aln_name] = {}
+ aln_pos = 0
+ _ = list(aln_dict[aln]["PATH.MATCH"].keys())
+ start_end_ids = _[0], _[-1]
+
+ for node_id, orient in aln_dict[aln_name]["PATH.MATCH"].items():
+ for path_name in nodes[node_id]["PATHS"].keys():
+ if node_id == start_end_ids[0] :
+ _apos = (aln_dict[aln_name]["QRY.START"], nodes_length[node_id])
+ _ppos = (
+ nodes[node_id]["PATHS"][path_name][0]+int(aln_dict[aln_name]["ALN.START"]),
+ nodes[node_id]["PATHS"][path_name][1]
+ )
+ elif node_id == start_end_ids[-1]:
+ _ppos = (
+ nodes[node_id]["PATHS"][path_name][0],
+ nodes[node_id]["PATHS"][path_name][0]+(int(aln_dict[aln_name]["ALN.BLOCK.LEN"]) - aln_pos)
+ )
+ _apos = (aln_pos, aln_dict[aln_name]["END"])
else :
- if in_aln or not len(cur_aln):
- cur_aln.append({
- "IN": False,
- "NODES": {
- node_id: strand
- }
- })
- in_aln = False
-
- else : cur_aln[-1]["NODES"][node_id] = strand
+ _ppos = nodes[node_id]["PATHS"][path_name]
+ #_apos = (aln_pos, aln_pos+nodes_length[node_id])
- aln_dict[aln_name]["HAP.MATCH"][path_name] = cur_aln.copy()
- print(f"{path_name} :", cur_aln)
+
+ _dict = {
+ "NODE.ID": node_id,
+ "P.ORIENT": paths[path_name]["NODES"][node_id],
+ "A.ORIENT": aln_dict[aln_name]["PATH.MATCH"][node_id],
+ "P.POS": _ppos,
+ #"A.POS": _apos,
+ "CIGAR": aln_dict[aln_name]["CIGAR"][node_id]
+ }
- # Traversing the HAP.MATCH to get alignment
- for
+ try :
+ GA[aln_name][path_name].append(_dict)
+ except :
+ GA[aln_name][path_name] = [_dict]
+
+print(GA)
+
+# Creating the final dictionnary called LA (Linear alignment):
+# Here we merged previous results to get full alignments
+for
--
GitLab
From 9c80010ecf0c5d79f0060653e9aaf5acd5f92319 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 17 May 2024 13:46:21 +0200
Subject: [PATCH 07/30] Update gaf2aln.py
---
gaf2aln.py | 48 ++++++++++++++++++++++++++++++++++++++++--------
1 file changed, 40 insertions(+), 8 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index afc61e2..e8f8406 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -156,7 +156,7 @@ nodes_length = {}
# Nodes dictionnary structured as follow :
# {<ALN.NODE.ID> : {PATHS: {<PATH.NAME>: (start, end)}}}
nodes = {
- node_id: {"PATHS": {}} for node_id in aln_nodes
+ node_id: {"PATHS": {}, "ALN": {}} for node_id in aln_nodes
}
# Paths dictionnary structured as follow :
# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
@@ -221,6 +221,29 @@ for path_name in paths.keys():
except :
cur_pos += nodes_length[path_node]+1
+print(f"[gaf2aln::Graph alignment processing] Computing nodes positions in each alignement...")
+# Adding nodes positions relative to path
+for aln_name in aln_dict.keys():
+ print(f"[gaf2aln::Graph processing] Running on {aln_name} ...")
+ cur_pos = 0
+ _ = list(aln_dict[aln_name]["PATH.MATCH"].keys())
+ start_end_id = (_[0], _[-1])
+ for node_id in aln_dict[aln_name]["PATH.MATCH"].keys():
+ if node_id == start_end_id[0]:
+ start_pos = int(aln_dict[aln_name]["ALN.START"])
+ end_pos = nodes_length[node_id]-int(aln_dict[aln_name]["ALN.START"])
+ elif node_id == start_end_id[1]:
+ start_pos = cur_pos
+ end_pos = int(aln_dict[aln_name]["ALN.END"])
+ else :
+ start_pos = cur_pos
+ end_pos = cur_pos+nodes_length[node_id]
+
+ nodes[node_id]["ALN"][aln_name] = (start_pos, end_pos)
+ cur_pos = end_pos+1
+
+print(nodes)
+
# Calculating CIGAR for each nodes in each aln
print(f"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...")
# Iterating over alignments
@@ -260,14 +283,12 @@ GA = {}
print(f"[gaf2aln::Graph alignment processing] Lifting alignments coordinates paths positions...")
for aln_name in aln_dict.keys():
GA[aln_name] = {}
- aln_pos = 0
_ = list(aln_dict[aln]["PATH.MATCH"].keys())
start_end_ids = _[0], _[-1]
for node_id, orient in aln_dict[aln_name]["PATH.MATCH"].items():
for path_name in nodes[node_id]["PATHS"].keys():
if node_id == start_end_ids[0] :
- _apos = (aln_dict[aln_name]["QRY.START"], nodes_length[node_id])
_ppos = (
nodes[node_id]["PATHS"][path_name][0]+int(aln_dict[aln_name]["ALN.START"]),
nodes[node_id]["PATHS"][path_name][1]
@@ -275,12 +296,10 @@ for aln_name in aln_dict.keys():
elif node_id == start_end_ids[-1]:
_ppos = (
nodes[node_id]["PATHS"][path_name][0],
- nodes[node_id]["PATHS"][path_name][0]+(int(aln_dict[aln_name]["ALN.BLOCK.LEN"]) - aln_pos)
+ nodes[node_id]["PATHS"][path_name][0]+(int(aln_dict[aln_name]["ALN.END"]) - nodes[node_id]["ALN"][aln_name][0])
)
- _apos = (aln_pos, aln_dict[aln_name]["END"])
else :
_ppos = nodes[node_id]["PATHS"][path_name]
- #_apos = (aln_pos, aln_pos+nodes_length[node_id])
_dict = {
@@ -288,7 +307,6 @@ for aln_name in aln_dict.keys():
"P.ORIENT": paths[path_name]["NODES"][node_id],
"A.ORIENT": aln_dict[aln_name]["PATH.MATCH"][node_id],
"P.POS": _ppos,
- #"A.POS": _apos,
"CIGAR": aln_dict[aln_name]["CIGAR"][node_id]
}
@@ -302,8 +320,22 @@ print(GA)
# Creating the final dictionnary called LA (Linear alignment):
# Here we merged previous results to get full alignments
-for
+for aln_name in GA.keys():
+ for path_name, node_list in GA[aln_name].items():
+
+ contiguity = False
+ orient = 1
+ alns = []
+
+ for node_data in node_list:
+
+ cur_orient = (node_data["P.ORIENT"] == node_data["A.ORIENT"])
+
+ if not contiguity :
+ alns.append(
+ {"Q.START": nodes[node_data["NODE.ID"]]["ALN"][aln_name]}
+ )
--
GitLab
From 662abb3a4d5c573fc1f965bce54729876d9e858b Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 17 May 2024 18:58:10 +0200
Subject: [PATCH 08/30] Update gaf2aln.py
---
gaf2aln.py | 224 +++++++++++++++++++++++++++++------------------------
1 file changed, 122 insertions(+), 102 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index e8f8406..e9850b3 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -78,12 +78,10 @@ def cigar2basealn(cigar):
For example : "345=" -> ["=", "=", ..., "="] of length 345.
"""
_ = re.findall(r'\d+\D', cigar)
- print(_)
final_cigar = []
for match in _:
final_cigar += [match[-1]]*int(match[:-1])
- print(final_cigar)
return final_cigar
def basealn2cigar(base_aln_list):
@@ -125,10 +123,10 @@ for line in range(len(gaf_lines)):
}
## Splitting "PATH.MATCH" into a list
- aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = {
- str(node_id[:-1]): node_id[-1]
+ aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = [
+ (str(node_id[:-1]), node_id[-1])
for node_id in walk2path(aln_dict[f"ALN_{line+1}"]["PATH.MATCH"])
- }
+ ]
## Adding CIGAR
aln_dict[f"ALN_{line+1}"]["RAW.CIGAR"] = line_content[-1]
@@ -140,7 +138,7 @@ for line in range(len(gaf_lines)):
aln_nodes = np.unique([
str(node_id)
for aln in aln_dict.keys()
- for node_id in aln_dict[aln]["PATH.MATCH"].keys()
+ for node_id, orient in aln_dict[aln]["PATH.MATCH"]
]).tolist()
del gaf_lines, gaf_col
@@ -154,10 +152,12 @@ with open(args.gfa, 'r') as file:
# {<NODE.ID>: <NODE.LENGTH>}
nodes_length = {}
# Nodes dictionnary structured as follow :
-# {<ALN.NODE.ID> : {PATHS: {<PATH.NAME>: (start, end)}}}
-nodes = {
- node_id: {"PATHS": {}, "ALN": {}} for node_id in aln_nodes
-}
+# { <ALN.NODE.ID> : {
+# <PATH.NAME>: {"START": start, "END": end, "STRAND": strand),
+# <ALN.NAME>: {"START": start, "END": end, "S.OFF": start.offset, "E.OFF": end.offset, "STRAND": strand, "CIGAR": CIGAR}
+# }
+# }
+nodes = {node_id: {} for node_id in aln_nodes}
# Paths dictionnary structured as follow :
# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
paths = {}
@@ -201,48 +201,84 @@ for line in gfa_lines:
del gfa_lines
-# Creating GA (Graph alignment) dictionnary storing given info :
-# {<ALN_ID> :
-# { <PATH.ID> : [
-# { NODE.ID, P.ORIENT, A.ORIENT, P.POS, A.POS, CG }
-# ]
-# }
-# }
-
print(f"[gaf2aln::Graph alignment processing] Computing nodes positions in each paths...")
# Adding nodes positions relative to path
for path_name in paths.keys():
print(f"[gaf2aln::Graph processing] Running on {path_name} ...")
cur_pos = 0
+
+ # Iterating over nodes in the path
for path_node in paths[path_name]["NODES"].keys():
+ # Instead of checking if the node is one interesting node, we try to add to the nodes dict
try :
- nodes[path_node]["PATHS"][path_name] = (cur_pos, cur_pos+nodes_length[path_node])
+ nodes[path_node][path_name] = {
+ "START": cur_pos, # Start position of the node start in the currrent path
+ "END": cur_pos+nodes_length[path_node], # End position of the node end in the current path
+ "STRAND": paths[path_name]["NODES"][node_id] # Orientation of the node in the current path
+ }
+
cur_pos += nodes_length[path_node]+1
except :
cur_pos += nodes_length[path_node]+1
print(f"[gaf2aln::Graph alignment processing] Computing nodes positions in each alignement...")
# Adding nodes positions relative to path
-for aln_name in aln_dict.keys():
- print(f"[gaf2aln::Graph processing] Running on {aln_name} ...")
+
+def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length):
+ # Initializing current position in query
cur_pos = 0
- _ = list(aln_dict[aln_name]["PATH.MATCH"].keys())
- start_end_id = (_[0], _[-1])
- for node_id in aln_dict[aln_name]["PATH.MATCH"].keys():
+
+ # Getting start and end node ids
+ start_end_id = (aln_dict[aln_name]["PATH.MATCH"][0][0], aln_dict[aln_name]["PATH.MATCH"][-1][0])
+
+ # Creating result dictionnary
+ res = {}
+
+ ## Iterating over node_ids from the given alignment
+ for node_id, orient in aln_dict[aln_name]["PATH.MATCH"]:
+ # Adding entry for current node
+ res[node_id] = {aln_name: {}}
+
+ # First node
if node_id == start_end_id[0]:
- start_pos = int(aln_dict[aln_name]["ALN.START"])
- end_pos = nodes_length[node_id]-int(aln_dict[aln_name]["ALN.START"])
+ start_pos = 0
+ s_off = int(aln_dict[aln_name]["ALN.START"])
+ end_pos = nodes_length[node_id]-s_off
+ e_off = 0
+ # End node
elif node_id == start_end_id[1]:
start_pos = cur_pos
- end_pos = int(aln_dict[aln_name]["ALN.END"])
+ s_off = 0
+ end_pos = int(aln_dict[aln_name]["QRY.END"])
+ e_off = nodes_length[node_id]-(end_pos-cur_pos)
+ # Node in between
else :
start_pos = cur_pos
+ s_off, e_off = 0, 0
end_pos = cur_pos+nodes_length[node_id]
- nodes[node_id]["ALN"][aln_name] = (start_pos, end_pos)
- cur_pos = end_pos+1
+ res[node_id] = {
+ "START": start_pos, # Start position on the query
+ "END": end_pos, # End position on the query
+ "S.OFF": s_off, # Offset between the start of the alignment and the node's start
+ "E.OFF": e_off, # Offset between the end of the alignment and the node's end
+ "STRAND": orient # Orientation of the node in the alignment
+ }
+
+ cur_pos = end_pos
+ print(start_pos, end_pos, s_off, e_off, orient, nodes_length[node_id], cur_pos)
-print(nodes)
+ return res
+
+# Storing alignement
+aln_processing = {}
+for aln_name in aln_dict.keys():
+ print(f"[gaf2aln::Graph processing] Running on {aln_name} ...")
+
+ _ = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)
+
+ for node_id, res in _.items():
+ nodes[node_id][aln_name] = res
# Calculating CIGAR for each nodes in each aln
print(f"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...")
@@ -252,90 +288,74 @@ for aln in aln_dict.keys():
print(f"[gaf2aln::CIGAR processing] Running on {aln} ...")
# Getting the list of base level alignement (["=", "X", ...] from "1=1X...")
raw_cigar = cigar2basealn(aln_dict[aln]["RAW.CIGAR"])
-
- cur_pos = 0
- print(len(raw_cigar))
CIGAR={}
- aln_nodes_id = list(aln_dict[aln]["PATH.MATCH"].keys())
- for node_id in aln_nodes_id:
-
- # Starting aln node
- if node_id == aln_nodes_id[0] :
- _cigar = basealn2cigar(raw_cigar[
- 0:nodes_length[node_id]-int(aln_dict[aln]["ALN.START"])
- ])
- cur_pos += nodes_length[node_id]-int(aln_dict[aln]["ALN.START"])
- # Last aln node
- elif node_id == aln_nodes_id[-1]:
- _cigar = basealn2cigar(raw_cigar[cur_pos:])
- else :
- _cigar = basealn2cigar(raw_cigar[cur_pos:cur_pos+nodes_length[node_id]])
- cur_pos += nodes_length[node_id]
- CIGAR[node_id] = _cigar
-
- aln_dict[aln]["CIGAR"] = CIGAR
- print(CIGAR)
+ for node_id, orient in aln_dict[aln]["PATH.MATCH"]:
+
+ _cigar = basealn2cigar(raw_cigar[
+ nodes[node_id][aln]["START"]:nodes[node_id][aln]["END"]
+ ])
+ nodes[node_id][aln]["CIGAR"] = _cigar
+ #print(_cigar, nodes[node_id][aln]["START"], nodes[node_id][aln]["END"])
+
+#print(nodes)
-GA = {}
+# Lifting graph alignements to haplotype alignements
-# Computing alignments nodes positions in paths
-print(f"[gaf2aln::Graph alignment processing] Lifting alignments coordinates paths positions...")
+ALNS = {}
for aln_name in aln_dict.keys():
- GA[aln_name] = {}
- _ = list(aln_dict[aln]["PATH.MATCH"].keys())
- start_end_ids = _[0], _[-1]
-
- for node_id, orient in aln_dict[aln_name]["PATH.MATCH"].items():
- for path_name in nodes[node_id]["PATHS"].keys():
- if node_id == start_end_ids[0] :
- _ppos = (
- nodes[node_id]["PATHS"][path_name][0]+int(aln_dict[aln_name]["ALN.START"]),
- nodes[node_id]["PATHS"][path_name][1]
- )
- elif node_id == start_end_ids[-1]:
- _ppos = (
- nodes[node_id]["PATHS"][path_name][0],
- nodes[node_id]["PATHS"][path_name][0]+(int(aln_dict[aln_name]["ALN.END"]) - nodes[node_id]["ALN"][aln_name][0])
- )
- else :
- _ppos = nodes[node_id]["PATHS"][path_name]
-
-
- _dict = {
- "NODE.ID": node_id,
- "P.ORIENT": paths[path_name]["NODES"][node_id],
- "A.ORIENT": aln_dict[aln_name]["PATH.MATCH"][node_id],
- "P.POS": _ppos,
- "CIGAR": aln_dict[aln_name]["CIGAR"][node_id]
- }
+
+ for path_name in paths.keys():
+ ALNS[(path_name, aln_name)] = []
+ _ = []
+ for node_id, orient in aln_dict[aln_name]["PATH.MATCH"].items():
+
+ n_info = nodes[node_id]
try :
- GA[aln_name][path_name].append(_dict)
- except :
- GA[aln_name][path_name] = [_dict]
-
-print(GA)
-
-# Creating the final dictionnary called LA (Linear alignment):
-# Here we merged previous results to get full alignments
+ if n_info[aln_name]["STRAND"] == n_info[path_name]["STRAND"] :
+ t_start = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
+ t_end = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
+ else :
+ t_end = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
+ t_start = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
+
+ q_start = n_info[aln_name]["START"]
+ q_end = n_info[aln_name]["END"]
+ _CG = n_info[aln_name]["CIGAR"]
+
+ # Non empty temporary list of aln and ending of the last block is the same as the start of the new node :
+ if len(_) and _[-1]["T.END"] == t_start and _[-1]["Q.END"] == q_start:
+ tmp_aln["Q.END"] = q_end
+ tmp_aln["T.END"] = t_end
+ tmp_aln["CG"] += _CG
+ elif len(_) and _[-1]["T.END"] == t_start: # Following on the target not on the query (i.e. Insertion)
+ tmp_aln["T.END"] = t_end
+ tmp_aln["CG"] += f"{nodes_length[node_id]}I"
+ elif len(_) and _[-1]["Q.END"] == q_start: # Following on the query, not on the target (i.e. Deletion)
+ tmp_aln["Q.END"] = q_end
+ tmps_aln["CG"] += f"{nodes_length[node_id]}D"
+ else : # Else, completely different
+ tmp_aln = {
+ "Q.START": q_start ,
+ "Q.END": q_end,
+ "T.START": t_start,
+ "T.END": t_end,
+ "CG": _CG,
+ }
+ except:
+ # Node is not in the path
+ tmp_aln
-for aln_name in GA.keys():
+
- for path_name, node_list in GA[aln_name].items():
- contiguity = False
- orient = 1
- alns = []
-
- for node_data in node_list:
- cur_orient = (node_data["P.ORIENT"] == node_data["A.ORIENT"])
+
+
+
+
- if not contiguity :
- alns.append(
- {"Q.START": nodes[node_data["NODE.ID"]]["ALN"][aln_name]}
- )
--
GitLab
From a12ec03c1c4e18dea839511ffcf2b6d893cead6b Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Fri, 17 May 2024 20:38:15 +0200
Subject: [PATCH 09/30] Update gaf2aln.py
---
gaf2aln.py | 22 +++++++++++++++-------
1 file changed, 15 insertions(+), 7 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index e9850b3..60794ea 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -309,9 +309,13 @@ for aln_name in aln_dict.keys():
ALNS[(path_name, aln_name)] = []
_ = []
- for node_id, orient in aln_dict[aln_name]["PATH.MATCH"].items():
+ for node_id, orient in aln_dict[aln_name]["PATH.MATCH"]:
n_info = nodes[node_id]
+ q_start = n_info[aln_name]["START"]
+ q_end = n_info[aln_name]["END"]
+ _CG = n_info[aln_name]["CIGAR"]
+
try :
if n_info[aln_name]["STRAND"] == n_info[path_name]["STRAND"] :
t_start = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
@@ -320,10 +324,6 @@ for aln_name in aln_dict.keys():
t_end = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
t_start = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
- q_start = n_info[aln_name]["START"]
- q_end = n_info[aln_name]["END"]
- _CG = n_info[aln_name]["CIGAR"]
-
# Non empty temporary list of aln and ending of the last block is the same as the start of the new node :
if len(_) and _[-1]["T.END"] == t_start and _[-1]["Q.END"] == q_start:
tmp_aln["Q.END"] = q_end
@@ -337,7 +337,7 @@ for aln_name in aln_dict.keys():
tmps_aln["CG"] += f"{nodes_length[node_id]}D"
else : # Else, completely different
tmp_aln = {
- "Q.START": q_start ,
+ "Q.START": q_start,
"Q.END": q_end,
"T.START": t_start,
"T.END": t_end,
@@ -345,7 +345,15 @@ for aln_name in aln_dict.keys():
}
except:
# Node is not in the path
- tmp_aln
+ tmp_aln = {
+ "Q.START": q_start,
+ "Q.END": q_end,
+ "T.START": -1,
+ "T.END": -1,
+ "CG": f"{nodes_length[node_id]}D"
+ }
+
+print(ALNS)
--
GitLab
From a7751c03ea016645940f4c6653c471c7805ca342 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Tue, 21 May 2024 17:07:23 +0200
Subject: [PATCH 10/30] Update gaf2aln.py
---
gaf2aln.py | 129 ++++++++++++++++++++++++++++++++++++++++++-----------
1 file changed, 104 insertions(+), 25 deletions(-)
diff --git a/gaf2aln.py b/gaf2aln.py
index 60794ea..7657046 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -167,6 +167,55 @@ links = {}
# Parsing the gfa
print(f"[gaf2aln::GFA Parser] Extracting nodes, paths and links ...")
+
+def GFA_parser(gfa_lines, nodes = nodes):
+ _links, _nodes, _nodes_length, paths = {}, {}, {}, {}
+ for line in gfa_lines:
+ line_content = line[:-1].split("\t")
+ line_id = line_content[0]
+
+ # Segment line
+ if line_id == "S" :
+
+ _nodes_length[str(line_content[1])] = len(line_content[2])
+
+ # Link line
+ elif line_id == "L":
+ try :
+ _links[str(line_content[1])][str(line_content[3])] = {
+ "FROM": str(line_content[2]),
+ "TO": str(line_content[4])
+ }
+
+ except :
+ _links[str(line_content[1])] = {
+ str(line_content[3]) : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
+ }
+
+ # Path line
+ elif line_id == "P":
+ _paths[str(line_content[1])] = {
+ "NODES": {
+ str(node_id[:-1]): str(node_id[-1])
+ for node_id in line_content[2].split(',')
+ },
+ "CIGAR": line_content[3]
+ }
+
+ return [_links, _nodes, _nodes_length, _paths]
+
+# splits = np.quantile(range(len(gfa_lines)+1), q= np.array(args.threads+1)/args.threads, method='higher').tolist()
+# res = []
+# for i in range(1, len(splits)):
+# res.append(executor.submit(GFA_parser, gfa_lines[splits[i-1]:splits[i]]))
+
+# for out in res:
+# results = out.result()
+
+# for link_id, link_info in results[0].items():
+# links[]
+
+
for line in gfa_lines:
line_content = line[:-1].split("\t")
line_id = line_content[0]
@@ -201,27 +250,44 @@ for line in gfa_lines:
del gfa_lines
-print(f"[gaf2aln::Graph alignment processing] Computing nodes positions in each paths...")
-# Adding nodes positions relative to path
-for path_name in paths.keys():
- print(f"[gaf2aln::Graph processing] Running on {path_name} ...")
+print(f"[gaf2aln::Graph position processing] Computing nodes positions in each paths...")
+def get_node_pos(path_name, nodes = nodes, paths = paths, nodes_length = nodes_length):
+ print(f"[gaf2aln::Graph position processing] Running on {path_name} ...")
cur_pos = 0
+ out = {}
# Iterating over nodes in the path
for path_node in paths[path_name]["NODES"].keys():
# Instead of checking if the node is one interesting node, we try to add to the nodes dict
- try :
- nodes[path_node][path_name] = {
+ if path_node in aln_nodes :
+ out[path_node] = {
"START": cur_pos, # Start position of the node start in the currrent path
"END": cur_pos+nodes_length[path_node], # End position of the node end in the current path
- "STRAND": paths[path_name]["NODES"][node_id] # Orientation of the node in the current path
+ "STRAND": paths[path_name]["NODES"][path_node] # Orientation of the node in the current path
}
cur_pos += nodes_length[path_node]+1
- except :
+ else :
cur_pos += nodes_length[path_node]+1
-print(f"[gaf2aln::Graph alignment processing] Computing nodes positions in each alignement...")
+ return out
+
+res = {}
+executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
+# Adding nodes positions relative to path
+for path_name in paths.keys():
+ res[path_name] = executor.submit(get_node_pos, path_name)
+
+executor.shutdown(wait=True)
+
+for path_name, out in res.items():
+ results = out.result()
+ for path_node, node_pos in results.items():
+ nodes[path_node][path_name] = node_pos
+
+del res
+
+print(f"[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...")
# Adding nodes positions relative to path
def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length):
@@ -271,14 +337,22 @@ def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length
return res
# Storing alignement
-aln_processing = {}
+res = {}
+executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
for aln_name in aln_dict.keys():
- print(f"[gaf2aln::Graph processing] Running on {aln_name} ...")
+ print(f"[gaf2aln::Alignment position processing] Running on {aln_name} ...")
- _ = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)
+ res[aln_name] = executor.submit(get_aln_node_info, aln_name)
+ #res[aln_name] = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)
+
+executor.shutdown(wait=True)
- for node_id, res in _.items():
- nodes[node_id][aln_name] = res
+for aln_name, node_info in res.items():
+ results = node_info.result()
+ for node_id, info in results.items():
+ nodes[node_id][aln_name] = info
+
+del res
# Calculating CIGAR for each nodes in each aln
print(f"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...")
@@ -306,7 +380,6 @@ ALNS = {}
for aln_name in aln_dict.keys():
for path_name in paths.keys():
- ALNS[(path_name, aln_name)] = []
_ = []
for node_id, orient in aln_dict[aln_name]["PATH.MATCH"]:
@@ -316,7 +389,10 @@ for aln_name in aln_dict.keys():
q_end = n_info[aln_name]["END"]
_CG = n_info[aln_name]["CIGAR"]
- try :
+ print(node_id, path_name, q_start, q_end)
+ if path_name in list(n_info.keys()):
+ print("\tIn path")
+
if n_info[aln_name]["STRAND"] == n_info[path_name]["STRAND"] :
t_start = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
t_end = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
@@ -324,6 +400,8 @@ for aln_name in aln_dict.keys():
t_end = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
t_start = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
+ print("\t", t_start, t_end)
+
# Non empty temporary list of aln and ending of the last block is the same as the start of the new node :
if len(_) and _[-1]["T.END"] == t_start and _[-1]["Q.END"] == q_start:
tmp_aln["Q.END"] = q_end
@@ -343,17 +421,18 @@ for aln_name in aln_dict.keys():
"T.END": t_end,
"CG": _CG,
}
- except:
+ print("\t", tmp_aln)
+
+ else :
+ print("\tNot in path")
# Node is not in the path
- tmp_aln = {
- "Q.START": q_start,
- "Q.END": q_end,
- "T.START": -1,
- "T.END": -1,
- "CG": f"{nodes_length[node_id]}D"
- }
-print(ALNS)
+ _.append(tmp_aln)
+ ALNS[(path_name, aln_name)] = _
+
+## Debug
+for elem in ALNS[("TO1000#1#chr03", "ALN_1")]:
+ print(elem)
--
GitLab
From 56990c58b0a19b66c42d98e3a5b62dd4a5c90d49 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 31 May 2024 13:42:58 +0200
Subject: [PATCH 11/30] Create Anchors2Path.py
---
Anchors2Path.py | 166 ++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 166 insertions(+)
create mode 100644 Anchors2Path.py
diff --git a/Anchors2Path.py b/Anchors2Path.py
new file mode 100644
index 0000000..e99b48a
--- /dev/null
+++ b/Anchors2Path.py
@@ -0,0 +1,166 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Anchors2Path
+Give anchors nodes ids with their relative positions in a given path.
+
+@author: alexis.mergez@inrae.fr
+@version: 0.1
+"""
+import re
+import argparse
+import os
+import numpy as np
+import time
+import pandas as pd
+from functools import reduce
+import concurrent.futures
+import gzip
+
+version = "0.1"
+
+## Argument parser
+arg_parser = argparse.ArgumentParser(description='Anchors2Path')
+arg_parser.add_argument(
+ "--gfa",
+ "-g",
+ dest = "gfa",
+ required = True,
+ help = "GFA file"
+ )
+arg_parser.add_argument(
+ "--output",
+ "-o",
+ dest = "output",
+ required = True,
+ help = "Output name"
+ )
+arg_parser.add_argument(
+ "--threads",
+ "-t",
+ dest = "threads",
+ required = False,
+ default = 1,
+ type = int,
+ help = "Number of threads"
+ )
+arg_parser.add_argument(
+ '--version',
+ '-v',
+ action="store_true",
+ dest = "version",
+ help = "Show version"
+)
+arg_parser.add_argument(
+ '--progress',
+ '-P',
+ action="store_true",
+ dest = "progress",
+ help = "Show progress to stdout"
+)
+arg_parser.add_argument(
+ '--pathname',
+ '-r',
+ dest = "pathname",
+ required = True,
+ help = "Pathname"
+)
+args = arg_parser.parse_args()
+
+# Printing version and exiting if required
+if args.version:
+ print(version)
+ os._exit(0)
+
+# Timing the script
+start_time = time.time()
+
+## Reading the gfa into a list
+# If not gzipped :
+if args.gfa[-2:] != "gz" :
+ with open(args.gfa, 'r') as file:
+ gfaLines = file.readlines()
+
+# If gzipped :
+else :
+ with gzip.open(args.gfa, 'r') as file:
+ gfaLines = [line.decode() for line in file.readlines()]
+
+# Progress message
+if args.progress: print(f"[GFAstats::{panname}] Parsing gfa file...")
+
+# Initializing dictionnaries
+Anchors = {}
+## {<NODE_ID>: (path_start, path_end)}
+nodes_length = {}
+path_nodes = {}
+## {<path_id>: <nodes_list>}
+
+for line in gfaLines[1:]:
+
+ # Skipping comment lines
+ if line[0] == "#":
+ lineType = "#"
+
+ # Reading 3 first columns of the current line
+ else :
+ lineType, uid, value = line[:-1].split('\t')[:3]
+
+ if lineType == "S": # Segments = Nodes
+ nodes_length[int(uid)] = len(value)
+
+ elif lineType == "P": # Paths
+
+ path_nodes[uid] = [k[:-1] for k in value.split(",")]
+
+if args.progress:
+ print(f"[GFAstats::{panname}] Parsed in {round(time.time() - start_time, 2)}s")
+
+# Getting the list of anchor nodes
+node_path_count = {}
+## {<NODE_ID>: <Number of path traversing this node>}
+# Computing number of path traversing each nodes
+for path_id, node_list in path_nodes.items():
+ for node_id in node_list:
+
+ try :
+ node_path_count[node_id] += 1
+ except :
+ node_path_count[node_id] = 1
+
+# Searching anchors
+n_path = len(list(path_nodes.keys()))
+
+for node_id, count in node_path_count.items():
+ if count == n_path :
+ Anchors[node_id] = []
+
+# Computing path position for each node of the path of interest
+current_pos = 0
+for node_id in path_nodes[args.pathname]:
+ _end = current_pos + nodes_length[node_id]
+
+ # Trying to add anchors path position if it is an anchor
+ try :
+ Anchors[node_id].append( (current_pos, _end) )
+ except:
+ pass
+
+ current_pos = _end
+
+# Transforming data into a table
+ID, START, END = [], [], []
+for node_id, positions in Anchors.items():
+ for start, end in positions:
+ ID.append(node_id)
+ START.append(start)
+ END.append(end)
+
+df = pd.DataFrame(data = {
+ "NODE_ID": ID,
+ "START": START,
+ "END": END
+})
+
+df.to_csv(args.output, sep="\t")
+
--
GitLab
From 1296dd803f594d2aeea67cb8cc92d9c007492922 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 31 May 2024 13:49:06 +0200
Subject: [PATCH 12/30] Update Anchors2Path.py
---
Anchors2Path.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index e99b48a..3006ef3 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -87,7 +87,7 @@ else :
gfaLines = [line.decode() for line in file.readlines()]
# Progress message
-if args.progress: print(f"[GFAstats::{panname}] Parsing gfa file...")
+if args.progress: print(f"[Anchors2Path] Parsing gfa file...")
# Initializing dictionnaries
Anchors = {}
@@ -114,7 +114,7 @@ for line in gfaLines[1:]:
path_nodes[uid] = [k[:-1] for k in value.split(",")]
if args.progress:
- print(f"[GFAstats::{panname}] Parsed in {round(time.time() - start_time, 2)}s")
+ print(f"[Anchors2Path] Parsed in {round(time.time() - start_time, 2)}s")
# Getting the list of anchor nodes
node_path_count = {}
--
GitLab
From c7d03713a9893274b3f3c106176bfc8bee9b2637 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 31 May 2024 13:52:15 +0200
Subject: [PATCH 13/30] Update Anchors2Path.py
---
Anchors2Path.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index 3006ef3..2e64c83 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -111,7 +111,7 @@ for line in gfaLines[1:]:
elif lineType == "P": # Paths
- path_nodes[uid] = [k[:-1] for k in value.split(",")]
+ path_nodes[uid] = [int(k[:-1]) for k in value.split(",")]
if args.progress:
print(f"[Anchors2Path] Parsed in {round(time.time() - start_time, 2)}s")
@@ -142,7 +142,7 @@ for node_id in path_nodes[args.pathname]:
# Trying to add anchors path position if it is an anchor
try :
- Anchors[node_id].append( (current_pos, _end) )
+ Anchors[int(node_id)].append( (current_pos, _end) )
except:
pass
--
GitLab
From 537598f921ec0f2a987f9554109802399fafccdb Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 31 May 2024 13:56:17 +0200
Subject: [PATCH 14/30] Update Anchors2Path.py
---
Anchors2Path.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index 2e64c83..3111c6f 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -162,5 +162,5 @@ df = pd.DataFrame(data = {
"END": END
})
-df.to_csv(args.output, sep="\t")
+df.to_csv(args.output, sep="\t", index = False)
--
GitLab
From 745eee2cd00e33eb46ce7410baab3f15fdb05ac1 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 13:20:47 +0200
Subject: [PATCH 15/30] Update
---
.ipynb_checkpoints/gaf2aln-checkpoint.ipynb | 1723 +++++++++++++
gaf2aln.ipynb | 2443 +++++++++++++++++++
gaf2aln.py | 38 +-
3 files changed, 4180 insertions(+), 24 deletions(-)
create mode 100644 .ipynb_checkpoints/gaf2aln-checkpoint.ipynb
create mode 100644 gaf2aln.ipynb
diff --git a/.ipynb_checkpoints/gaf2aln-checkpoint.ipynb b/.ipynb_checkpoints/gaf2aln-checkpoint.ipynb
new file mode 100644
index 0000000..effb776
--- /dev/null
+++ b/.ipynb_checkpoints/gaf2aln-checkpoint.ipynb
@@ -0,0 +1,1723 @@
+{
+ "cells": [
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "id": "4ffaf9f6-cc1e-4190-9351-5431c930d25b",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import numpy as np\n",
+ "import pandas as pd\n",
+ "import argparse\n",
+ "import concurrent.futures\n",
+ "import os\n",
+ "import re\n",
+ "\n",
+ "# Replace for argparse arguments\n",
+ "class arguments():\n",
+ " gfa = \"/home/amergez/Documents/Scratch/LeChou/35Bra-v2a/35Bra-v2a.chr03.gfa\"\n",
+ " gaf = \"/home/amergez/Documents/Scratch/LeChou/35Bra-v2a/Mapping2Graph/GA.FLC2.aln.gaf\"\n",
+ " threads = 8\n",
+ " version = False\n",
+ "args = arguments()"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 2,
+ "id": "280c8847-22e8-4063-bde8-3e4e72cf20e7",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# Toolbox\n",
+ "def walk2path(walk):\n",
+ " \"\"\"\n",
+ " Takes a walk in a single string and returns a list of nodes id with signs (gfa v1 like)\n",
+ " \"\"\"\n",
+ " _ = re.findall(r'>\\w+|<\\w+', walk)\n",
+ " # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'\n",
+ " return [f'{elem[1:]}{(elem[0] == \">\")*\"+\"+(elem[0] == \"<\")*\"-\"}' for elem in _]\n",
+ "\n",
+ "def cigar2basealn(cigar):\n",
+ " \"\"\"\n",
+ " Takes a CIGAR string and convert it into a list of base level alignment.\n",
+ " For example : \"345=\" -> [\"=\", \"=\", ..., \"=\"] of length 345.\n",
+ " \"\"\"\n",
+ " _ = re.findall(r'\\d+\\D', cigar)\n",
+ " final_cigar = []\n",
+ " for match in _:\n",
+ " final_cigar += [match[-1]]*int(match[:-1])\n",
+ "\n",
+ " return final_cigar\n",
+ "\n",
+ "def basealn2cigar(base_aln_list):\n",
+ " \n",
+ " last_elem = base_aln_list[0]\n",
+ " CIGAR = [[1, last_elem]]\n",
+ " for elem in base_aln_list[1:]:\n",
+ " if elem == last_elem:\n",
+ " CIGAR[-1][0] += 1\n",
+ "\n",
+ " else :\n",
+ " CIGAR[-1][0] = str(CIGAR[-1][0])\n",
+ " CIGAR.append([1, elem])\n",
+ " last_elem = elem\n",
+ " CIGAR[-1][0] = str(CIGAR[-1][0])\n",
+ " return \"\".join([\"\".join(block) for block in CIGAR if block[1] != \"\"])"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 3,
+ "id": "be12e9d4-de76-4c8b-af84-6567549483f4",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::GAF Parser] Reading /home/amergez/Documents/Scratch/LeChou/35Bra-v2a/Mapping2Graph/GA.FLC2.aln.gaf ...\n",
+ "[gaf2aln::GAF Parser] Extracting alignments ...\n",
+ "{'ALN_1': {'QRY.NAME': 'FLC2.TO1000#1#chr03', 'QRY.LEN': '3735', 'QRY.START': '0', 'QRY.END': '3735', 'STRAND': '+', 'PATH.MATCH': [('7046526', '+'), ('7046528', '+'), ('7046530', '+'), ('7046531', '+'), ('7046532', '+'), ('7046533', '+'), ('7046534', '+'), ('7046536', '+'), ('7046537', '+'), ('7046539', '+'), ('7046541', '+'), ('7046542', '+'), ('7046544', '+'), ('7046546', '+'), ('7046547', '+'), ('7046549', '+'), ('7046551', '+'), ('7046552', '+'), ('7046554', '+'), ('7046556', '+'), ('7046556', '+'), ('7046556', '+'), ('7046557', '+'), ('7046558', '+'), ('7046559', '+'), ('7046560', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046562', '+'), ('7046564', '+'), ('7046565', '+'), ('7046567', '+'), ('7046568', '+'), ('7046570', '+'), ('7046571', '+'), ('7046573', '+'), ('7046574', '+'), ('7046576', '+'), ('7046577', '+'), ('7046579', '+'), ('7046581', '+'), ('7046583', '+'), ('7046584', '+'), ('7046586', '+'), ('7046587', '+'), ('7046589', '+'), ('7046590', '+'), ('7046592', '+'), ('7046593', '+'), ('7046594', '+'), ('7046596', '+'), ('7046597', '+'), ('7046599', '+'), ('7046600', '+'), ('7046601', '+'), ('7046603', '+'), ('7046604', '+'), ('7046606', '+'), ('7046608', '+'), ('7046609', '+'), ('7046621', '+'), ('7046622', '+'), ('7046624', '+'), ('7046625', '+'), ('7046626', '+'), ('7046628', '+'), ('7046631', '+'), ('7046673', '+'), ('7046631', '+'), ('7046673', '+'), ('7046631', '+'), ('7046632', '+'), ('7046634', '+'), ('7046635', '+'), ('7046637', '+'), ('7046638', '+'), ('7046639', '+'), ('7046641', '+'), ('7046644', '+'), ('7046646', '+'), ('7046647', '+'), ('7046649', '+'), ('7046650', '+'), ('7046652', '+'), ('7046653', '+'), ('7046654', '+'), ('7046656', '+'), ('7046657', '+'), ('7046659', '+'), ('7046660', '+'), ('7046662', '+'), ('7046663', '+'), ('7046665', '+'), ('7046667', '+'), ('7046668', '+'), ('7046670', '+'), ('7046671', '+'), ('7046674', '+'), ('7046675', '+'), ('7046674', '+'), ('7046675', '+'), ('7046676', '+'), ('7046678', '+'), ('7046679', '+'), ('7046680', '+'), ('7046682', '+'), ('7046684', '+'), ('7046685', '+'), ('7046686', '+'), ('7046688', '+'), ('7046690', '+'), ('7046692', '+'), ('7046693', '+'), ('7046695', '+'), ('7046696', '+'), ('7046698', '+'), ('7046700', '+'), ('7046702', '+'), ('7046703', '+'), ('7046704', '+'), ('7046706', '+'), ('7046707', '+'), ('7046709', '+'), ('7046710', '+'), ('7046712', '+'), ('7046713', '+'), ('7046715', '+'), ('7046718', '+'), ('7046717', '+'), ('7046718', '+'), ('7046717', '+'), ('7046718', '+'), ('7046720', '+'), ('7046722', '+'), ('7046724', '+'), ('7046725', '+'), ('7046727', '+'), ('7046728', '+'), ('7046729', '+'), ('7046730', '+'), ('7046731', '+'), ('7046733', '+'), ('7046735', '+'), ('7046736', '+'), ('7046738', '+'), ('7046739', '+'), ('7046740', '+'), ('7046738', '+'), ('7046739', '+'), ('7046740', '+'), ('7046738', '+'), ('7046739', '+'), ('7046741', '+')], 'PATH.LEN': '3822', 'ALN.START': '77', 'ALN.END': '3812', 'RES.MATCH': '3735', 'ALN.BLOCK.LEN': '3735', 'MAPPING.QUAL': '60', 'RAW.CIGAR': 'cg:Z:3735=', 'TAGS': 'AS:f:3735,dv:f:0,id:f:1'}, 'ALN_2': {'QRY.NAME': 'FLC2.TO1000#1#chr03', 'QRY.LEN': '3735', 'QRY.START': '0', 'QRY.END': '3735', 'STRAND': '+', 'PATH.MATCH': [('7594382', '+'), ('7594369', '+'), ('7594371', '+'), ('7594021', '+'), ('7594286', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594374', '+'), ('7594375', '+'), ('7594626', '+'), ('7594011', '+'), ('7594374', '+'), ('7594375', '+'), ('7594369', '+'), ('7594371', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594241', '+'), ('7594248', '+'), ('7594286', '+'), ('7594311', '+'), ('7594315', '+'), ('7594311', '+'), ('7594330', '+'), ('7594311', '+'), ('7594315', '+'), ('7594374', '+'), ('7594311', '+'), ('7594374', '+'), ('7594369', '+'), ('7594021', '+'), ('7594026', '+'), ('7594021', '+'), ('7594021', '+'), ('7594026', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594286', '+'), ('7594374', '+'), ('7594021', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594315', '+'), ('7594286', '+'), ('7594311', '+'), ('7594374', '+'), ('7594021', '+'), ('7594286', '+'), ('7594286', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594374', '+'), ('7594375', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594375', '+'), ('7594374', '+'), ('7594350', '+'), ('7594264', '+'), ('7594207', '+'), ('7594225', '+'), ('7594227', '+'), ('7594120', '+'), ('7594132', '+'), ('7594165', '+'), ('7594172', '+')], 'PATH.LEN': '61224', 'ALN.START': '0', 'ALN.END': '3735', 'RES.MATCH': '3734', 'ALN.BLOCK.LEN': '3735', 'MAPPING.QUAL': '0', 'RAW.CIGAR': 'cg:Z:57=1X3677=', 'TAGS': 'AS:f:3732.06,dv:f:0.000267738,id:f:0.999732'}}\n"
+ ]
+ }
+ ],
+ "source": [
+ "# Parsing the .gaf file\n",
+ "print(f\"[gaf2aln::GAF Parser] Reading {args.gaf} ...\")\n",
+ "with open(args.gaf, 'r') as file:\n",
+ " gaf_lines = file.readlines()\n",
+ "\n",
+ "gaf_col = [\n",
+ " \"QRY.NAME\", \"QRY.LEN\", \"QRY.START\", \"QRY.END\", \"STRAND\", \n",
+ " \"PATH.MATCH\", \"PATH.LEN\", \"ALN.START\", \"ALN.END\",\n",
+ " \"RES.MATCH\", \"ALN.BLOCK.LEN\", \"MAPPING.QUAL\"\n",
+ " ]\n",
+ "\n",
+ "# Creating dictionnary to store alignments\n",
+ "print(f\"[gaf2aln::GAF Parser] Extracting alignments ...\")\n",
+ "aln_dict = {}\n",
+ "for line in range(len(gaf_lines)):\n",
+ " ## Splitting the line by tabulation\n",
+ " line_content = gaf_lines[line][:-1].split('\\t')\n",
+ "\n",
+ " ## Adding alignement info to dictionnary\n",
+ " aln_dict[f\"ALN_{line+1}\"] = {\n",
+ " gaf_col[i]: line_content[i] for i in range(len(gaf_col))\n",
+ " }\n",
+ " \n",
+ " ## Splitting \"PATH.MATCH\" into a list\n",
+ " aln_dict[f\"ALN_{line+1}\"][\"PATH.MATCH\"] = [\n",
+ " (str(node_id[:-1]), node_id[-1]) \n",
+ " for node_id in walk2path(aln_dict[f\"ALN_{line+1}\"][\"PATH.MATCH\"])\n",
+ " ]\n",
+ "\n",
+ " ## Adding CIGAR\n",
+ " aln_dict[f\"ALN_{line+1}\"][\"RAW.CIGAR\"] = line_content[-1]\n",
+ "\n",
+ " ## Adding tags\n",
+ " aln_dict[f\"ALN_{line+1}\"][\"TAGS\"] = \",\".join(line_content[13:-1])\n",
+ "\n",
+ "# Getting nodes of interest ids\n",
+ "aln_nodes = np.unique([\n",
+ " str(node_id) \n",
+ " for aln in aln_dict.keys() \n",
+ " for node_id, orient in aln_dict[aln][\"PATH.MATCH\"]\n",
+ "]).tolist()\n",
+ "\n",
+ "print(aln_dict)\n",
+ "del gaf_lines, gaf_col"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 4,
+ "id": "2f891424-0d88-4fd3-99ff-b0a8c90587ff",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::GFA Parser] Reading /home/amergez/Documents/Scratch/LeChou/35Bra-v2a/35Bra-v2a.chr03.gfa ...\n",
+ "[gaf2aln::GFA Parser] Extracting nodes, paths and links ...\n"
+ ]
+ }
+ ],
+ "source": [
+ "# Parsing the .gfa\n",
+ "print(f\"[gaf2aln::GFA Parser] Reading {args.gfa} ...\")\n",
+ "with open(args.gfa, 'r') as file:\n",
+ " gfa_lines = file.readlines()\n",
+ "\n",
+ "# Nodes length dictionnary structured as follow :\n",
+ "# {<NODE.ID>: <NODE.LENGTH>}\n",
+ "nodes_length = {}\n",
+ "# Nodes dictionnary structured as follow :\n",
+ "# { <ALN.NODE.ID> : {\n",
+ "# <PATH.NAME>: {\"START\": start, \"END\": end, \"STRAND\": strand), \n",
+ "# <ALN.NAME>: {\"START\": start, \"END\": end, \"S.OFF\": start.offset, \"E.OFF\": end.offset, \"STRAND\": strand, \"CIGAR\": CIGAR}\n",
+ "# }\n",
+ "# }\n",
+ "nodes = {node_id: {} for node_id in aln_nodes}\n",
+ "# Paths dictionnary structured as follow :\n",
+ "# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}\n",
+ "paths = {}\n",
+ "# Links dictionnary structured as follow : \n",
+ "# {<FROM.NODE.ID>: {<TO.NODE.ID>: {FROM.ORIENT: <FROM.ORIENT>, TO.ORIENT: <TO.ORIENT>}}}\n",
+ "links = {}\n",
+ "\n",
+ "# Parsing the gfa\n",
+ "print(f\"[gaf2aln::GFA Parser] Extracting nodes, paths and links ...\")\n",
+ "\n",
+ "def GFA_parser(gfa_lines, nodes = nodes):\n",
+ " _links, _nodes, _nodes_length, paths = {}, {}, {}, {}\n",
+ " for line in gfa_lines:\n",
+ " line_content = line[:-1].split(\"\\t\")\n",
+ " line_id = line_content[0]\n",
+ " \n",
+ " # Segment line\n",
+ " if line_id == \"S\" :\n",
+ " \n",
+ " _nodes_length[str(line_content[1])] = len(line_content[2])\n",
+ " \n",
+ " # Link line\n",
+ " elif line_id == \"L\":\n",
+ " try :\n",
+ " _links[str(line_content[1])][str(line_content[3])] = {\n",
+ " \"FROM\": str(line_content[2]), \n",
+ " \"TO\": str(line_content[4])\n",
+ " }\n",
+ "\n",
+ " except :\n",
+ " _links[str(line_content[1])] = {\n",
+ " str(line_content[3]) : {\"FROM.ORIENT\": str(line_content[2]), \"TO.ORIENT\": str(line_content[4])}\n",
+ " }\n",
+ "\n",
+ " # Path line\n",
+ " elif line_id == \"P\":\n",
+ " _paths[str(line_content[1])] = {\n",
+ " \"NODES\": {\n",
+ " str(node_id[:-1]): str(node_id[-1])\n",
+ " for node_id in line_content[2].split(',')\n",
+ " },\n",
+ " \"CIGAR\": line_content[3]\n",
+ " }\n",
+ "\n",
+ " return [_links, _nodes, _nodes_length, _paths]\n",
+ "\n",
+ "# splits = np.quantile(range(len(gfa_lines)+1), q= np.array(args.threads+1)/args.threads, method='higher').tolist()\n",
+ "# res = []\n",
+ "# for i in range(1, len(splits)):\n",
+ "# res.append(executor.submit(GFA_parser, gfa_lines[splits[i-1]:splits[i]]))\n",
+ "\n",
+ "# for out in res:\n",
+ "# results = out.result()\n",
+ "\n",
+ "# for link_id, link_info in results[0].items():\n",
+ "# links[]\n",
+ "\n",
+ "\n",
+ "for line in gfa_lines:\n",
+ " line_content = line[:-1].split(\"\\t\")\n",
+ " line_id = line_content[0]\n",
+ " \n",
+ " # Segment line\n",
+ " if line_id == \"S\" :\n",
+ " \n",
+ " nodes_length[str(line_content[1])] = len(line_content[2])\n",
+ " \n",
+ " # Link line\n",
+ " elif line_id == \"L\":\n",
+ " try :\n",
+ " links[str(line_content[1])][str(line_content[3])] = {\n",
+ " \"FROM\": str(line_content[2]), \n",
+ " \"TO\": str(line_content[4])\n",
+ " }\n",
+ "\n",
+ " except :\n",
+ " links[str(line_content[1])] = {\n",
+ " str(line_content[3]) : {\"FROM.ORIENT\": str(line_content[2]), \"TO.ORIENT\": str(line_content[4])}\n",
+ " }\n",
+ "\n",
+ " # Path line\n",
+ " elif line_id == \"P\":\n",
+ " paths[str(line_content[1])] = {\n",
+ " \"NODES\": {\n",
+ " str(node_id[:-1]): str(node_id[-1])\n",
+ " for node_id in line_content[2].split(',')\n",
+ " },\n",
+ " \"CIGAR\": line_content[3]\n",
+ " }\n",
+ "\n",
+ "del gfa_lines"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 5,
+ "id": "a403c88e-54ea-4a67-9047-dc44eba7f51a",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::Graph position processing] Computing nodes positions in each paths...\n",
+ "[gaf2aln::Graph position processing] Running on Capitata#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on D101#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on D134#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on G06-09-28#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on G07-DH-33#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on HDEM#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on Korso#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on M249#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on OX-heart#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on PL021#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on RC34#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T02#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T03#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T04#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T06#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T07#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T08#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T09#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T10#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T11#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T12#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T13#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T14#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T15#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T16#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T17#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T18#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T19#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T21#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T24#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T25#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T26#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T27#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on TO1000#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on W1701#1#chr03 ...\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(f\"[gaf2aln::Graph position processing] Computing nodes positions in each paths...\")\n",
+ "def get_node_pos(path_name, nodes = nodes, paths = paths, nodes_length = nodes_length):\n",
+ " print(f\"[gaf2aln::Graph position processing] Running on {path_name} ...\")\n",
+ " cur_pos = 0\n",
+ "\n",
+ " out = {}\n",
+ " # Iterating over nodes in the path\n",
+ " for path_node in paths[path_name][\"NODES\"].keys():\n",
+ " # Instead of checking if the node is one interesting node, we try to add to the nodes dict\n",
+ " if path_node in aln_nodes :\n",
+ " out[path_node] = {\n",
+ " \"START\": cur_pos, # Start position of the node start in the currrent path\n",
+ " \"END\": cur_pos+nodes_length[path_node], # End position of the node end in the current path\n",
+ " \"STRAND\": paths[path_name][\"NODES\"][path_node] # Orientation of the node in the current path\n",
+ " } \n",
+ "\n",
+ " cur_pos += nodes_length[path_node]+1\n",
+ " else :\n",
+ " cur_pos += nodes_length[path_node]+1\n",
+ "\n",
+ " return out\n",
+ "\n",
+ "res = {}\n",
+ "executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)\n",
+ "# Adding nodes positions relative to path\n",
+ "for path_name in paths.keys():\n",
+ " res[path_name] = executor.submit(get_node_pos, path_name)\n",
+ "\n",
+ "executor.shutdown(wait=True)\n",
+ "\n",
+ "for path_name, out in res.items():\n",
+ " results = out.result()\n",
+ " for path_node, node_pos in results.items():\n",
+ " nodes[path_node][path_name] = node_pos\n",
+ "\n",
+ "del res"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 6,
+ "id": "bed36bd5-30eb-4d02-8b52-1ae5d753f8f8",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...\n",
+ "[gaf2aln::Alignment position processing] Running on ALN_1 ...\n",
+ "0 77 77 0 + 154 77\n",
+ "77 82 0 0 + 5 82\n",
+ "82 83 0 0 + 1 83\n",
+ "83 138 0 0 + 55 138\n",
+ "138 139 0 0 + 1 139\n",
+ "139 202 0 0 + 63 202\n",
+ "202 203 0 0 + 1 203\n",
+ "203 379 0 0 + 176 379\n",
+ "379 380 0 0 + 1 380\n",
+ "380 429 0 0 + 49 429\n",
+ "429 430 0 0 + 1 430\n",
+ "430 457 0 0 + 27 457\n",
+ "457 492 0 0 + 35 492\n",
+ "492 494 0 0 + 2 494\n",
+ "494 497 0 0 + 3 497\n",
+ "497 507 0 0 + 10 507\n",
+ "507 508 0 0 + 1 508\n",
+ "508 564 0 0 + 56 564\n",
+ "564 566 0 0 + 2 566\n",
+ "566 567 0 0 + 1 567\n",
+ "567 568 0 0 + 1 568\n",
+ "568 569 0 0 + 1 569\n",
+ "569 824 0 0 + 255 824\n",
+ "824 826 0 0 + 2 826\n",
+ "826 858 0 0 + 32 858\n",
+ "858 859 0 0 + 1 859\n",
+ "859 860 0 0 + 1 860\n",
+ "860 861 0 0 + 1 861\n",
+ "861 862 0 0 + 1 862\n",
+ "862 863 0 0 + 1 863\n",
+ "863 864 0 0 + 1 864\n",
+ "864 865 0 0 + 1 865\n",
+ "865 866 0 0 + 1 866\n",
+ "866 867 0 0 + 1 867\n",
+ "867 868 0 0 + 1 868\n",
+ "868 869 0 0 + 1 869\n",
+ "869 913 0 0 + 44 913\n",
+ "913 919 0 0 + 6 919\n",
+ "919 978 0 0 + 59 978\n",
+ "978 979 0 0 + 1 979\n",
+ "979 1038 0 0 + 59 1038\n",
+ "1038 1045 0 0 + 7 1045\n",
+ "1045 1046 0 0 + 1 1046\n",
+ "1046 1080 0 0 + 34 1080\n",
+ "1080 1081 0 0 + 1 1081\n",
+ "1081 1107 0 0 + 26 1107\n",
+ "1107 1108 0 0 + 1 1108\n",
+ "1108 1183 0 0 + 75 1183\n",
+ "1183 1186 0 0 + 3 1186\n",
+ "1186 1224 0 0 + 38 1224\n",
+ "1224 1257 0 0 + 33 1257\n",
+ "1257 1289 0 0 + 32 1289\n",
+ "1289 1311 0 0 + 22 1311\n",
+ "1311 1359 0 0 + 48 1359\n",
+ "1359 1382 0 0 + 23 1382\n",
+ "1382 1434 0 0 + 52 1434\n",
+ "1434 1451 0 0 + 17 1451\n",
+ "1451 1531 0 0 + 80 1531\n",
+ "1531 1532 0 0 + 1 1532\n",
+ "1532 1543 0 0 + 11 1543\n",
+ "1543 1544 0 0 + 1 1544\n",
+ "1544 1572 0 0 + 28 1572\n",
+ "1572 1573 0 0 + 1 1573\n",
+ "1573 1587 0 0 + 14 1587\n",
+ "1587 1588 0 0 + 1 1588\n",
+ "1588 1616 0 0 + 28 1616\n",
+ "1616 1617 0 0 + 1 1617\n",
+ "1617 1646 0 0 + 29 1646\n",
+ "1646 1661 0 0 + 15 1661\n",
+ "1661 1673 0 0 + 12 1673\n",
+ "1673 1674 0 0 + 1 1674\n",
+ "1674 1726 0 0 + 52 1726\n",
+ "1726 1727 0 0 + 1 1727\n",
+ "1727 1762 0 0 + 35 1762\n",
+ "1762 1763 0 0 + 1 1763\n",
+ "1763 1764 0 0 + 1 1764\n",
+ "1764 1765 0 0 + 1 1765\n",
+ "1765 1766 0 0 + 1 1766\n",
+ "1766 1767 0 0 + 1 1767\n",
+ "1767 1824 0 0 + 57 1824\n",
+ "1824 1825 0 0 + 1 1825\n",
+ "1825 1975 0 0 + 150 1975\n",
+ "1975 1976 0 0 + 1 1976\n",
+ "1976 2015 0 0 + 39 2015\n",
+ "2015 2016 0 0 + 1 2016\n",
+ "2016 2047 0 0 + 31 2047\n",
+ "2047 2055 0 0 + 8 2055\n",
+ "2055 2056 0 0 + 1 2056\n",
+ "2056 2120 0 0 + 64 2120\n",
+ "2120 2121 0 0 + 1 2121\n",
+ "2121 2157 0 0 + 36 2157\n",
+ "2157 2158 0 0 + 1 2158\n",
+ "2158 2170 0 0 + 12 2170\n",
+ "2170 2171 0 0 + 1 2171\n",
+ "2171 2205 0 0 + 34 2205\n",
+ "2205 2206 0 0 + 1 2206\n",
+ "2206 2344 0 0 + 138 2344\n",
+ "2344 2345 0 0 + 1 2345\n",
+ "2345 2364 0 0 + 19 2364\n",
+ "2364 2383 0 0 + 19 2383\n",
+ "2383 2408 0 0 + 25 2408\n",
+ "2408 2409 0 0 + 1 2409\n",
+ "2409 2441 0 0 + 32 2441\n",
+ "2441 2442 0 0 + 1 2442\n",
+ "2442 2580 0 0 + 138 2580\n",
+ "2580 2581 0 0 + 1 2581\n",
+ "2581 2582 0 0 + 1 2582\n",
+ "2582 2583 0 0 + 1 2583\n",
+ "2583 2584 0 0 + 1 2584\n",
+ "2584 2764 0 0 + 180 2764\n",
+ "2764 2765 0 0 + 1 2765\n",
+ "2765 2797 0 0 + 32 2797\n",
+ "2797 2798 0 0 + 1 2798\n",
+ "2798 2878 0 0 + 80 2878\n",
+ "2878 2879 0 0 + 1 2879\n",
+ "2879 2951 0 0 + 72 2951\n",
+ "2951 2952 0 0 + 1 2952\n",
+ "2952 3002 0 0 + 50 3002\n",
+ "3002 3077 0 0 + 75 3077\n",
+ "3077 3078 0 0 + 1 3078\n",
+ "3078 3093 0 0 + 15 3093\n",
+ "3093 3094 0 0 + 1 3094\n",
+ "3094 3097 0 0 + 3 3097\n",
+ "3097 3140 0 0 + 43 3140\n",
+ "3140 3210 0 0 + 70 3210\n",
+ "3210 3211 0 0 + 1 3211\n",
+ "3211 3229 0 0 + 18 3229\n",
+ "3229 3230 0 0 + 1 3230\n",
+ "3230 3276 0 0 + 46 3276\n",
+ "3276 3277 0 0 + 1 3277\n",
+ "3277 3315 0 0 + 38 3315\n",
+ "3315 3316 0 0 + 1 3316\n",
+ "3316 3322 0 0 + 6 3322\n",
+ "3322 3323 0 0 + 1 3323\n",
+ "3323 3348 0 0 + 25 3348\n",
+ "3348 3349 0 0 + 1 3349\n",
+ "3349 3350 0 0 + 1 3350\n",
+ "3350 3351 0 0 + 1 3351\n",
+ "3351 3352 0 0 + 1 3352\n",
+ "3352 3353 0 0 + 1 3353\n",
+ "3353 3354 0 0 + 1 3354\n",
+ "3354 3356 0 0 + 2 3356\n",
+ "3356 3357 0 0 + 1[gaf2aln::Alignment position processing] Running on ALN_2 ...\n",
+ " 03357 \n",
+ "13357 03489 0 +0 10 1+\n",
+ " 1132 23489 \n",
+ "03489 03490 +0 10 2\n",
+ "+2 3 10 34900\n",
+ " 3490+ 36421 03 \n",
+ "03 +4 0 1520 3642+\n",
+ " 36421 36444 \n",
+ "04 05 +0 20 3644+\n",
+ " 36441 5 \n",
+ "36855 06 00 +0 41+ 36851\n",
+ " 36856 \n",
+ "36876 07 00 +0 2+ 36871\n",
+ " 36877 \n",
+ "36937 08 00 +0 6+ 36931\n",
+ " 36938 \n",
+ "36948 09 00 +0 1+ 36941\n",
+ " 36949 \n",
+ "37089 010 00 +0 14+ 37081\n",
+ " 370810 \n",
+ "370910 011 00 +0 1+ 37091\n",
+ " 370911 \n",
+ "371011 012 00 +0 1+ 37101\n",
+ " 371012 \n",
+ "371412 013 00 +0 4+ 37141\n",
+ " 371413 \n",
+ "371513 014 00 +0 1+ 37151\n",
+ " 371514 \n",
+ "371614 015 00 +0 1+ 37161\n",
+ " 371615 \n",
+ "372015 016 00 +0 4+ 37201\n",
+ " 372016\n",
+ " 163721 170 00 0+ +1 13721 \n",
+ "173721\n",
+ " 173722 180 00 0+ +1 13722 \n",
+ "183722\n",
+ " 183735 19 00 100 ++ 231 373519\n",
+ "\n",
+ "19 20 0 0 + 1 20\n",
+ "20 21 0 0 + 1 21\n",
+ "21 22 0 0 + 1 22\n",
+ "22 23 0 0 + 1 23\n",
+ "23 24 0 0 + 1 24\n",
+ "24 25 0 0 + 1 25\n",
+ "25 26 0 0 + 1 26\n",
+ "26 27 0 0 + 1 27\n",
+ "27 28 0 0 + 1 28\n",
+ "28 29 0 0 + 1 29\n",
+ "29 30 0 0 + 1 30\n",
+ "30 31 0 0 + 1 31\n",
+ "31 32 0 0 + 1 32\n",
+ "32 33 0 0 + 1 33\n",
+ "33 34 0 0 + 1 34\n",
+ "34 35 0 0 + 1 35\n",
+ "35 36 0 0 + 1 36\n",
+ "36 37 0 0 + 1 37\n",
+ "37 38 0 0 + 1 38\n",
+ "38 39 0 0 + 1 39\n",
+ "39 40 0 0 + 1 40\n",
+ "40 41 0 0 + 1 41\n",
+ "41 42 0 0 + 1 42\n",
+ "42 43 0 0 + 1 43\n",
+ "43 44 0 0 + 1 44\n",
+ "44 45 0 0 + 1 45\n",
+ "45 46 0 0 + 1 46\n",
+ "46 47 0 0 + 1 47\n",
+ "47 48 0 0 + 1 48\n",
+ "48 49 0 0 + 1 49\n",
+ "49 50 0 0 + 1 50\n",
+ "50 51 0 0 + 1 51\n",
+ "51 52 0 0 + 1 52\n",
+ "52 53 0 0 + 1 53\n",
+ "53 54 0 0 + 1 54\n",
+ "54 55 0 0 + 1 55\n",
+ "55 56 0 0 + 1 56\n",
+ "56 57 0 0 + 1 57\n",
+ "57 58 0 0 + 1 58\n",
+ "58 59 0 0 + 1 59\n",
+ "59 60 0 0 + 1 60\n",
+ "60 61 0 0 + 1 61\n",
+ "61 62 0 0 + 1 62\n",
+ "62 63 0 0 + 1 63\n",
+ "63 64 0 0 + 1 64\n",
+ "64 65 0 0 + 1 65\n",
+ "65 66 0 0 + 1 66\n",
+ "66 67 0 0 + 1 67\n",
+ "67 68 0 0 + 1 68\n",
+ "68 69 0 0 + 1 69\n",
+ "69 70 0 0 + 1 70\n",
+ "70 71 0 0 + 1 71\n",
+ "71 72 0 0 + 1 72\n",
+ "72 73 0 0 + 1 73\n",
+ "73 74 0 0 + 1 74\n",
+ "74 75 0 0 + 1 75\n",
+ "75 76 0 0 + 1 76\n",
+ "76 77 0 0 + 1 77\n",
+ "77 78 0 0 + 1 78\n",
+ "78 3735 0 57489 + 61146 3735\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(f\"[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...\")\n",
+ "# Adding nodes positions relative to path\n",
+ "\n",
+ "def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length):\n",
+ " # Initializing current position in query\n",
+ " cur_pos = 0\n",
+ "\n",
+ " # Getting start and end node ids\n",
+ " start_end_id = (aln_dict[aln_name][\"PATH.MATCH\"][0][0], aln_dict[aln_name][\"PATH.MATCH\"][-1][0])\n",
+ "\n",
+ " # Creating result dictionnary\n",
+ " res = {}\n",
+ "\n",
+ " ## Iterating over node_ids from the given alignment\n",
+ " for node_id, orient in aln_dict[aln_name][\"PATH.MATCH\"]:\n",
+ " # Adding entry for current node\n",
+ " res[node_id] = {aln_name: {}}\n",
+ "\n",
+ " # First node\n",
+ " if node_id == start_end_id[0]:\n",
+ " start_pos = 0\n",
+ " s_off = int(aln_dict[aln_name][\"ALN.START\"])\n",
+ " end_pos = nodes_length[node_id]-s_off\n",
+ " e_off = 0\n",
+ " # End node\n",
+ " elif node_id == start_end_id[1]:\n",
+ " start_pos = cur_pos\n",
+ " s_off = 0\n",
+ " end_pos = int(aln_dict[aln_name][\"QRY.END\"])\n",
+ " e_off = nodes_length[node_id]-(end_pos-cur_pos)\n",
+ " # Node in between\n",
+ " else :\n",
+ " start_pos = cur_pos\n",
+ " s_off, e_off = 0, 0\n",
+ " end_pos = cur_pos+nodes_length[node_id]\n",
+ "\n",
+ " res[node_id] = {\n",
+ " \"START\": start_pos, # Start position on the query\n",
+ " \"END\": end_pos, # End position on the query\n",
+ " \"S.OFF\": s_off, # Offset between the start of the alignment and the node's start\n",
+ " \"E.OFF\": e_off, # Offset between the end of the alignment and the node's end \n",
+ " \"STRAND\": orient # Orientation of the node in the alignment\n",
+ " }\n",
+ " \n",
+ " cur_pos = end_pos\n",
+ " print(start_pos, end_pos, s_off, e_off, orient, nodes_length[node_id], cur_pos)\n",
+ "\n",
+ " return res\n",
+ "\n",
+ "# Storing alignement \n",
+ "res = {}\n",
+ "executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)\n",
+ "for aln_name in aln_dict.keys():\n",
+ " print(f\"[gaf2aln::Alignment position processing] Running on {aln_name} ...\")\n",
+ " \n",
+ " res[aln_name] = executor.submit(get_aln_node_info, aln_name)\n",
+ " #res[aln_name] = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)\n",
+ "\n",
+ "executor.shutdown(wait=True)\n",
+ "\n",
+ "for aln_name, node_info in res.items():\n",
+ " results = node_info.result()\n",
+ " for node_id, info in results.items():\n",
+ " nodes[node_id][aln_name] = info\n",
+ "\n",
+ "del res"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 7,
+ "id": "4c30727c-7ffc-4852-ad81-ca2a5a7f9957",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...\n",
+ "[gaf2aln::CIGAR processing] Running on ALN_1 ...\n",
+ "[gaf2aln::CIGAR processing] Running on ALN_2 ...\n"
+ ]
+ }
+ ],
+ "source": [
+ "# Calculating CIGAR for each nodes in each aln\n",
+ "print(f\"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...\")\n",
+ "# Iterating over alignments\n",
+ "for aln in aln_dict.keys():\n",
+ " \n",
+ " print(f\"[gaf2aln::CIGAR processing] Running on {aln} ...\")\n",
+ " # Getting the list of base level alignement ([\"=\", \"X\", ...] from \"1=1X...\")\n",
+ " raw_cigar = cigar2basealn(aln_dict[aln][\"RAW.CIGAR\"])\n",
+ " CIGAR={}\n",
+ "\n",
+ " for node_id, orient in aln_dict[aln][\"PATH.MATCH\"]:\n",
+ "\n",
+ " _cigar = basealn2cigar(raw_cigar[\n",
+ " nodes[node_id][aln][\"START\"]:nodes[node_id][aln][\"END\"]\n",
+ " ])\n",
+ " nodes[node_id][aln][\"CIGAR\"] = _cigar\n",
+ " #print(_cigar, nodes[node_id][aln][\"START\"], nodes[node_id][aln][\"END\"])"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 8,
+ "id": "e15e4762-cd71-4afe-bc74-ebe44869fee6",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "ALN_1\n",
+ "7046526 D134#1#chr03 0 77\n",
+ "\tIn path\n",
+ "\t 73306158 73306235\n",
+ "skipped\n",
+ "\n",
+ "7046528 D134#1#chr03 77 82\n",
+ "\tIn path\n",
+ "\t 73306238 73306243\n",
+ "{'Q.START': 0, 'Q.END': 77, 'T.START': 73306158, 'T.END': 73306235, 'CG': '77='}\n",
+ "7046530 D134#1#chr03 82 83\n",
+ "\tNot in path\n",
+ "7046531 D134#1#chr03 83 138\n",
+ "\tIn path\n",
+ "\t 73306246 73306301\n",
+ "{'Q.START': 77, 'Q.END': 82, 'T.START': 73306238, 'T.END': 73306243, 'CG': '5='}\n",
+ "7046532 D134#1#chr03 138 139\n",
+ "\tNot in path\n",
+ "7046533 D134#1#chr03 139 202\n",
+ "\tIn path\n",
+ "\t 73306302 73306365\n",
+ "{'Q.START': 83, 'Q.END': 138, 'T.START': 73306246, 'T.END': 73306301, 'CG': '55='}\n",
+ "7046534 D134#1#chr03 202 203\n",
+ "\tIn path\n",
+ "\t 73306366 73306367\n",
+ "{'Q.START': 139, 'Q.END': 202, 'T.START': 73306302, 'T.END': 73306365, 'CG': '63='}\n",
+ "7046536 D134#1#chr03 203 379\n",
+ "\tIn path\n",
+ "\t 73306368 73306544\n",
+ "{'Q.START': 202, 'Q.END': 203, 'T.START': 73306366, 'T.END': 73306367, 'CG': '1='}\n",
+ "7046537 D134#1#chr03 379 380\n",
+ "\tIn path\n",
+ "\t 73306545 73306546\n",
+ "{'Q.START': 203, 'Q.END': 379, 'T.START': 73306368, 'T.END': 73306544, 'CG': '176='}\n",
+ "7046539 D134#1#chr03 380 429\n",
+ "\tIn path\n",
+ "\t 73306547 73306596\n",
+ "{'Q.START': 379, 'Q.END': 380, 'T.START': 73306545, 'T.END': 73306546, 'CG': '1='}\n",
+ "7046541 D134#1#chr03 429 430\n",
+ "\tIn path\n",
+ "\t 73306597 73306598\n",
+ "{'Q.START': 380, 'Q.END': 429, 'T.START': 73306547, 'T.END': 73306596, 'CG': '49='}\n",
+ "7046542 D134#1#chr03 430 457\n",
+ "\tIn path\n",
+ "\t 73306599 73306626\n",
+ "{'Q.START': 429, 'Q.END': 430, 'T.START': 73306597, 'T.END': 73306598, 'CG': '1='}\n",
+ "7046544 D134#1#chr03 457 492\n",
+ "\tIn path\n",
+ "\t 73306641 73306676\n",
+ "{'Q.START': 430, 'Q.END': 457, 'T.START': 73306599, 'T.END': 73306626, 'CG': '27='}\n",
+ "7046546 D134#1#chr03 492 494\n",
+ "\tNot in path\n",
+ "7046547 D134#1#chr03 494 497\n",
+ "\tNot in path\n",
+ "7046549 D134#1#chr03 497 507\n",
+ "\tNot in path\n",
+ "7046551 D134#1#chr03 507 508\n",
+ "\tNot in path\n",
+ "7046552 D134#1#chr03 508 564\n",
+ "\tIn path\n",
+ "\t 73306694 73306750\n",
+ "{'Q.START': 457, 'Q.END': 492, 'T.START': 73306641, 'T.END': 73306676, 'CG': '35='}\n",
+ "7046554 D134#1#chr03 564 566\n",
+ "\tNot in path\n",
+ "7046556 D134#1#chr03 568 569\n",
+ "\tIn path\n",
+ "\t 73306753 73306754\n",
+ "{'Q.START': 508, 'Q.END': 564, 'T.START': 73306694, 'T.END': 73306750, 'CG': '56='}\n",
+ "7046556 D134#1#chr03 568 569\n",
+ "\tIn path\n",
+ "\t 73306753 73306754\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
+ "7046556 D134#1#chr03 568 569\n",
+ "\tIn path\n",
+ "\t 73306753 73306754\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
+ "7046557 D134#1#chr03 569 824\n",
+ "\tIn path\n",
+ "\t 73306755 73307010\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
+ "7046558 D134#1#chr03 824 826\n",
+ "\tNot in path\n",
+ "7046559 D134#1#chr03 826 858\n",
+ "\tIn path\n",
+ "\t 73307011 73307043\n",
+ "{'Q.START': 569, 'Q.END': 824, 'T.START': 73306755, 'T.END': 73307010, 'CG': '255='}\n",
+ "7046560 D134#1#chr03 858 859\n",
+ "\tIn path\n",
+ "\t 73307044 73307045\n",
+ "{'Q.START': 826, 'Q.END': 858, 'T.START': 73307011, 'T.END': 73307043, 'CG': '32='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 858, 'Q.END': 859, 'T.START': 73307044, 'T.END': 73307045, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046562 D134#1#chr03 869 913\n",
+ "\tIn path\n",
+ "\t 73307048 73307092\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046564 D134#1#chr03 913 919\n",
+ "\tIn path\n",
+ "\t 73307093 73307099\n",
+ "{'Q.START': 869, 'Q.END': 913, 'T.START': 73307048, 'T.END': 73307092, 'CG': '44='}\n",
+ "7046565 D134#1#chr03 919 978\n",
+ "\tIn path\n",
+ "\t 73307100 73307159\n",
+ "{'Q.START': 913, 'Q.END': 919, 'T.START': 73307093, 'T.END': 73307099, 'CG': '6='}\n",
+ "7046567 D134#1#chr03 978 979\n",
+ "\tIn path\n",
+ "\t 73307160 73307161\n",
+ "{'Q.START': 919, 'Q.END': 978, 'T.START': 73307100, 'T.END': 73307159, 'CG': '59='}\n",
+ "7046568 D134#1#chr03 979 1038\n",
+ "\tIn path\n",
+ "\t 73307162 73307221\n",
+ "{'Q.START': 978, 'Q.END': 979, 'T.START': 73307160, 'T.END': 73307161, 'CG': '1='}\n",
+ "7046570 D134#1#chr03 1038 1045\n",
+ "\tIn path\n",
+ "\t 73307224 73307231\n",
+ "{'Q.START': 979, 'Q.END': 1038, 'T.START': 73307162, 'T.END': 73307221, 'CG': '59='}\n",
+ "7046571 D134#1#chr03 1045 1046\n",
+ "\tIn path\n",
+ "\t 73307232 73307233\n",
+ "{'Q.START': 1038, 'Q.END': 1045, 'T.START': 73307224, 'T.END': 73307231, 'CG': '7='}\n",
+ "7046573 D134#1#chr03 1046 1080\n",
+ "\tIn path\n",
+ "\t 73307234 73307268\n",
+ "{'Q.START': 1045, 'Q.END': 1046, 'T.START': 73307232, 'T.END': 73307233, 'CG': '1='}\n",
+ "7046574 D134#1#chr03 1080 1081\n",
+ "\tIn path\n",
+ "\t 73307269 73307270\n",
+ "{'Q.START': 1046, 'Q.END': 1080, 'T.START': 73307234, 'T.END': 73307268, 'CG': '34='}\n",
+ "7046576 D134#1#chr03 1081 1107\n",
+ "\tIn path\n",
+ "\t 73307271 73307297\n",
+ "{'Q.START': 1080, 'Q.END': 1081, 'T.START': 73307269, 'T.END': 73307270, 'CG': '1='}\n",
+ "7046577 D134#1#chr03 1107 1108\n",
+ "\tNot in path\n",
+ "7046579 D134#1#chr03 1108 1183\n",
+ "\tIn path\n",
+ "\t 73307300 73307375\n",
+ "{'Q.START': 1081, 'Q.END': 1107, 'T.START': 73307271, 'T.END': 73307297, 'CG': '26='}\n",
+ "7046581 D134#1#chr03 1183 1186\n",
+ "\tIn path\n",
+ "\t 73307376 73307379\n",
+ "{'Q.START': 1108, 'Q.END': 1183, 'T.START': 73307300, 'T.END': 73307375, 'CG': '75='}\n",
+ "7046583 D134#1#chr03 1186 1224\n",
+ "\tNot in path\n",
+ "7046584 D134#1#chr03 1224 1257\n",
+ "\tIn path\n",
+ "\t 73307419 73307452\n",
+ "{'Q.START': 1183, 'Q.END': 1186, 'T.START': 73307376, 'T.END': 73307379, 'CG': '3='}\n",
+ "7046586 D134#1#chr03 1257 1289\n",
+ "\tNot in path\n",
+ "7046587 D134#1#chr03 1289 1311\n",
+ "\tIn path\n",
+ "\t 73307475 73307497\n",
+ "{'Q.START': 1224, 'Q.END': 1257, 'T.START': 73307419, 'T.END': 73307452, 'CG': '33='}\n",
+ "7046589 D134#1#chr03 1311 1359\n",
+ "\tNot in path\n",
+ "7046590 D134#1#chr03 1359 1382\n",
+ "\tIn path\n",
+ "\t 73307546 73307569\n",
+ "{'Q.START': 1289, 'Q.END': 1311, 'T.START': 73307475, 'T.END': 73307497, 'CG': '22='}\n",
+ "7046592 D134#1#chr03 1382 1434\n",
+ "\tNot in path\n",
+ "7046593 D134#1#chr03 1434 1451\n",
+ "\tIn path\n",
+ "\t 73307643 73307660\n",
+ "{'Q.START': 1359, 'Q.END': 1382, 'T.START': 73307546, 'T.END': 73307569, 'CG': '23='}\n",
+ "7046594 D134#1#chr03 1451 1531\n",
+ "\tIn path\n",
+ "\t 73307661 73307741\n",
+ "{'Q.START': 1434, 'Q.END': 1451, 'T.START': 73307643, 'T.END': 73307660, 'CG': '17='}\n",
+ "7046596 D134#1#chr03 1531 1532\n",
+ "\tNot in path\n",
+ "7046597 D134#1#chr03 1532 1543\n",
+ "\tIn path\n",
+ "\t 73307744 73307755\n",
+ "{'Q.START': 1451, 'Q.END': 1531, 'T.START': 73307661, 'T.END': 73307741, 'CG': '80='}\n",
+ "7046599 D134#1#chr03 1543 1544\n",
+ "\tNot in path\n",
+ "7046600 D134#1#chr03 1544 1572\n",
+ "\tIn path\n",
+ "\t 73307758 73307786\n",
+ "{'Q.START': 1532, 'Q.END': 1543, 'T.START': 73307744, 'T.END': 73307755, 'CG': '11='}\n",
+ "7046601 D134#1#chr03 1572 1573\n",
+ "\tIn path\n",
+ "\t 73307787 73307788\n",
+ "{'Q.START': 1544, 'Q.END': 1572, 'T.START': 73307758, 'T.END': 73307786, 'CG': '28='}\n",
+ "7046603 D134#1#chr03 1573 1587\n",
+ "\tIn path\n",
+ "\t 73307789 73307803\n",
+ "{'Q.START': 1572, 'Q.END': 1573, 'T.START': 73307787, 'T.END': 73307788, 'CG': '1='}\n",
+ "7046604 D134#1#chr03 1587 1588\n",
+ "\tNot in path\n",
+ "7046606 D134#1#chr03 1588 1616\n",
+ "\tIn path\n",
+ "\t 73307806 73307834\n",
+ "{'Q.START': 1573, 'Q.END': 1587, 'T.START': 73307789, 'T.END': 73307803, 'CG': '14='}\n",
+ "7046608 D134#1#chr03 1616 1617\n",
+ "\tIn path\n",
+ "\t 73307835 73307836\n",
+ "{'Q.START': 1588, 'Q.END': 1616, 'T.START': 73307806, 'T.END': 73307834, 'CG': '28='}\n",
+ "7046609 D134#1#chr03 1617 1646\n",
+ "\tIn path\n",
+ "\t 73307837 73307866\n",
+ "{'Q.START': 1616, 'Q.END': 1617, 'T.START': 73307835, 'T.END': 73307836, 'CG': '1='}\n",
+ "7046621 D134#1#chr03 1646 1661\n",
+ "\tIn path\n",
+ "\t 73307867 73307882\n",
+ "{'Q.START': 1617, 'Q.END': 1646, 'T.START': 73307837, 'T.END': 73307866, 'CG': '29='}\n",
+ "7046622 D134#1#chr03 1661 1673\n",
+ "\tIn path\n",
+ "\t 73307883 73307895\n",
+ "{'Q.START': 1646, 'Q.END': 1661, 'T.START': 73307867, 'T.END': 73307882, 'CG': '15='}\n",
+ "7046624 D134#1#chr03 1673 1674\n",
+ "\tIn path\n",
+ "\t 73307896 73307897\n",
+ "{'Q.START': 1661, 'Q.END': 1673, 'T.START': 73307883, 'T.END': 73307895, 'CG': '12='}\n",
+ "7046625 D134#1#chr03 1674 1726\n",
+ "\tIn path\n",
+ "\t 73307898 73307950\n",
+ "{'Q.START': 1673, 'Q.END': 1674, 'T.START': 73307896, 'T.END': 73307897, 'CG': '1='}\n",
+ "7046626 D134#1#chr03 1726 1727\n",
+ "\tNot in path\n",
+ "7046628 D134#1#chr03 1727 1762\n",
+ "\tIn path\n",
+ "\t 73307953 73307988\n",
+ "{'Q.START': 1674, 'Q.END': 1726, 'T.START': 73307898, 'T.END': 73307950, 'CG': '52='}\n",
+ "7046631 D134#1#chr03 1766 1767\n",
+ "\tIn path\n",
+ "\t 73307991 73307992\n",
+ "{'Q.START': 1727, 'Q.END': 1762, 'T.START': 73307953, 'T.END': 73307988, 'CG': '35='}\n",
+ "7046673 D134#1#chr03 1765 1766\n",
+ "\tIn path\n",
+ "\t 73307993 73307994\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
+ "7046631 D134#1#chr03 1766 1767\n",
+ "\tIn path\n",
+ "\t 73307991 73307992\n",
+ "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}\n",
+ "7046673 D134#1#chr03 1765 1766\n",
+ "\tIn path\n",
+ "\t 73307993 73307994\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
+ "7046631 D134#1#chr03 1766 1767\n",
+ "\tIn path\n",
+ "\t 73307991 73307992\n",
+ "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}\n",
+ "7046632 D134#1#chr03 1767 1824\n",
+ "\tIn path\n",
+ "\t 73307995 73308052\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
+ "7046634 D134#1#chr03 1824 1825\n",
+ "\tIn path\n",
+ "\t 73308053 73308054\n",
+ "{'Q.START': 1767, 'Q.END': 1824, 'T.START': 73307995, 'T.END': 73308052, 'CG': '57='}\n",
+ "7046635 D134#1#chr03 1825 1975\n",
+ "\tIn path\n",
+ "\t 73308055 73308205\n",
+ "{'Q.START': 1824, 'Q.END': 1825, 'T.START': 73308053, 'T.END': 73308054, 'CG': '1='}\n",
+ "7046637 D134#1#chr03 1975 1976\n",
+ "\tNot in path\n",
+ "7046638 D134#1#chr03 1976 2015\n",
+ "\tIn path\n",
+ "\t 73308208 73308247\n",
+ "{'Q.START': 1825, 'Q.END': 1975, 'T.START': 73308055, 'T.END': 73308205, 'CG': '150='}\n",
+ "7046639 D134#1#chr03 2015 2016\n",
+ "\tNot in path\n",
+ "7046641 D134#1#chr03 2016 2047\n",
+ "\tIn path\n",
+ "\t 73308250 73308281\n",
+ "{'Q.START': 1976, 'Q.END': 2015, 'T.START': 73308208, 'T.END': 73308247, 'CG': '39='}\n",
+ "7046644 D134#1#chr03 2047 2055\n",
+ "\tIn path\n",
+ "\t 73308286 73308294\n",
+ "{'Q.START': 2016, 'Q.END': 2047, 'T.START': 73308250, 'T.END': 73308281, 'CG': '31='}\n",
+ "7046646 D134#1#chr03 2055 2056\n",
+ "\tNot in path\n",
+ "7046647 D134#1#chr03 2056 2120\n",
+ "\tIn path\n",
+ "\t 73308297 73308361\n",
+ "{'Q.START': 2047, 'Q.END': 2055, 'T.START': 73308286, 'T.END': 73308294, 'CG': '8='}\n",
+ "7046649 D134#1#chr03 2120 2121\n",
+ "\tIn path\n",
+ "\t 73308362 73308363\n",
+ "{'Q.START': 2056, 'Q.END': 2120, 'T.START': 73308297, 'T.END': 73308361, 'CG': '64='}\n",
+ "7046650 D134#1#chr03 2121 2157\n",
+ "\tIn path\n",
+ "\t 73308364 73308400\n",
+ "{'Q.START': 2120, 'Q.END': 2121, 'T.START': 73308362, 'T.END': 73308363, 'CG': '1='}\n",
+ "7046652 D134#1#chr03 2157 2158\n",
+ "\tNot in path\n",
+ "7046653 D134#1#chr03 2158 2170\n",
+ "\tIn path\n",
+ "\t 73308403 73308415\n",
+ "{'Q.START': 2121, 'Q.END': 2157, 'T.START': 73308364, 'T.END': 73308400, 'CG': '36='}\n",
+ "7046654 D134#1#chr03 2170 2171\n",
+ "\tIn path\n",
+ "\t 73308416 73308417\n",
+ "{'Q.START': 2158, 'Q.END': 2170, 'T.START': 73308403, 'T.END': 73308415, 'CG': '12='}\n",
+ "7046656 D134#1#chr03 2171 2205\n",
+ "\tIn path\n",
+ "\t 73308418 73308452\n",
+ "{'Q.START': 2170, 'Q.END': 2171, 'T.START': 73308416, 'T.END': 73308417, 'CG': '1='}\n",
+ "7046657 D134#1#chr03 2205 2206\n",
+ "\tNot in path\n",
+ "7046659 D134#1#chr03 2206 2344\n",
+ "\tIn path\n",
+ "\t 73308455 73308593\n",
+ "{'Q.START': 2171, 'Q.END': 2205, 'T.START': 73308418, 'T.END': 73308452, 'CG': '34='}\n",
+ "7046660 D134#1#chr03 2344 2345\n",
+ "\tNot in path\n",
+ "7046662 D134#1#chr03 2345 2364\n",
+ "\tIn path\n",
+ "\t 73308596 73308615\n",
+ "{'Q.START': 2206, 'Q.END': 2344, 'T.START': 73308455, 'T.END': 73308593, 'CG': '138='}\n",
+ "7046663 D134#1#chr03 2364 2383\n",
+ "\tIn path\n",
+ "\t 73308616 73308635\n",
+ "{'Q.START': 2345, 'Q.END': 2364, 'T.START': 73308596, 'T.END': 73308615, 'CG': '19='}\n",
+ "7046665 D134#1#chr03 2383 2408\n",
+ "\tIn path\n",
+ "\t 73308636 73308661\n",
+ "{'Q.START': 2364, 'Q.END': 2383, 'T.START': 73308616, 'T.END': 73308635, 'CG': '19='}\n",
+ "7046667 D134#1#chr03 2408 2409\n",
+ "\tIn path\n",
+ "\t 73308662 73308663\n",
+ "{'Q.START': 2383, 'Q.END': 2408, 'T.START': 73308636, 'T.END': 73308661, 'CG': '25='}\n",
+ "7046668 D134#1#chr03 2409 2441\n",
+ "\tIn path\n",
+ "\t 73308664 73308696\n",
+ "{'Q.START': 2408, 'Q.END': 2409, 'T.START': 73308662, 'T.END': 73308663, 'CG': '1='}\n",
+ "7046670 D134#1#chr03 2441 2442\n",
+ "\tIn path\n",
+ "\t 73308697 73308698\n",
+ "{'Q.START': 2409, 'Q.END': 2441, 'T.START': 73308664, 'T.END': 73308696, 'CG': '32='}\n",
+ "7046671 D134#1#chr03 2442 2580\n",
+ "\tIn path\n",
+ "\t 73308699 73308837\n",
+ "{'Q.START': 2441, 'Q.END': 2442, 'T.START': 73308697, 'T.END': 73308698, 'CG': '1='}\n",
+ "7046674 D134#1#chr03 2582 2583\n",
+ "\tIn path\n",
+ "\t 73308838 73308839\n",
+ "{'Q.START': 2442, 'Q.END': 2580, 'T.START': 73308699, 'T.END': 73308837, 'CG': '138='}\n",
+ "7046675 D134#1#chr03 2583 2584\n",
+ "\tIn path\n",
+ "\t 73308840 73308841\n",
+ "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}\n",
+ "7046674 D134#1#chr03 2582 2583\n",
+ "\tIn path\n",
+ "\t 73308838 73308839\n",
+ "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}\n",
+ "7046675 D134#1#chr03 2583 2584\n",
+ "\tIn path\n",
+ "\t 73308840 73308841\n",
+ "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}\n",
+ "7046676 D134#1#chr03 2584 2764\n",
+ "\tIn path\n",
+ "\t 73308842 73309022\n",
+ "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}\n",
+ "7046678 D134#1#chr03 2764 2765\n",
+ "\tNot in path\n",
+ "7046679 D134#1#chr03 2765 2797\n",
+ "\tIn path\n",
+ "\t 73309025 73309057\n",
+ "{'Q.START': 2584, 'Q.END': 2764, 'T.START': 73308842, 'T.END': 73309022, 'CG': '180='}\n",
+ "7046680 D134#1#chr03 2797 2798\n",
+ "\tNot in path\n",
+ "7046682 D134#1#chr03 2798 2878\n",
+ "\tIn path\n",
+ "\t 73309060 73309140\n",
+ "{'Q.START': 2765, 'Q.END': 2797, 'T.START': 73309025, 'T.END': 73309057, 'CG': '32='}\n",
+ "7046684 D134#1#chr03 2878 2879\n",
+ "\tIn path\n",
+ "\t 73309141 73309142\n",
+ "{'Q.START': 2798, 'Q.END': 2878, 'T.START': 73309060, 'T.END': 73309140, 'CG': '80='}\n",
+ "7046685 D134#1#chr03 2879 2951\n",
+ "\tIn path\n",
+ "\t 73309143 73309215\n",
+ "{'Q.START': 2878, 'Q.END': 2879, 'T.START': 73309141, 'T.END': 73309142, 'CG': '1='}\n",
+ "7046686 D134#1#chr03 2951 2952\n",
+ "\tIn path\n",
+ "\t 73309216 73309217\n",
+ "{'Q.START': 2879, 'Q.END': 2951, 'T.START': 73309143, 'T.END': 73309215, 'CG': '72='}\n",
+ "7046688 D134#1#chr03 2952 3002\n",
+ "\tIn path\n",
+ "\t 73309218 73309268\n",
+ "{'Q.START': 2951, 'Q.END': 2952, 'T.START': 73309216, 'T.END': 73309217, 'CG': '1='}\n",
+ "7046690 D134#1#chr03 3002 3077\n",
+ "\tIn path\n",
+ "\t 73309271 73309346\n",
+ "{'Q.START': 2952, 'Q.END': 3002, 'T.START': 73309218, 'T.END': 73309268, 'CG': '50='}\n",
+ "7046692 D134#1#chr03 3077 3078\n",
+ "\tIn path\n",
+ "\t 73309347 73309348\n",
+ "{'Q.START': 3002, 'Q.END': 3077, 'T.START': 73309271, 'T.END': 73309346, 'CG': '75='}\n",
+ "7046693 D134#1#chr03 3078 3093\n",
+ "\tIn path\n",
+ "\t 73309349 73309364\n",
+ "{'Q.START': 3077, 'Q.END': 3078, 'T.START': 73309347, 'T.END': 73309348, 'CG': '1='}\n",
+ "7046695 D134#1#chr03 3093 3094\n",
+ "\tNot in path\n",
+ "7046696 D134#1#chr03 3094 3097\n",
+ "\tIn path\n",
+ "\t 73309367 73309370\n",
+ "{'Q.START': 3078, 'Q.END': 3093, 'T.START': 73309349, 'T.END': 73309364, 'CG': '15='}\n",
+ "7046698 D134#1#chr03 3097 3140\n",
+ "\tIn path\n",
+ "\t 73309371 73309414\n",
+ "{'Q.START': 3094, 'Q.END': 3097, 'T.START': 73309367, 'T.END': 73309370, 'CG': '3='}\n",
+ "7046700 D134#1#chr03 3140 3210\n",
+ "\tIn path\n",
+ "\t 73309415 73309485\n",
+ "{'Q.START': 3097, 'Q.END': 3140, 'T.START': 73309371, 'T.END': 73309414, 'CG': '43='}\n",
+ "7046702 D134#1#chr03 3210 3211\n",
+ "\tIn path\n",
+ "\t 73309486 73309487\n",
+ "{'Q.START': 3140, 'Q.END': 3210, 'T.START': 73309415, 'T.END': 73309485, 'CG': '70='}\n",
+ "7046703 D134#1#chr03 3211 3229\n",
+ "\tIn path\n",
+ "\t 73309488 73309506\n",
+ "{'Q.START': 3210, 'Q.END': 3211, 'T.START': 73309486, 'T.END': 73309487, 'CG': '1='}\n",
+ "7046704 D134#1#chr03 3229 3230\n",
+ "\tIn path\n",
+ "\t 73309507 73309508\n",
+ "{'Q.START': 3211, 'Q.END': 3229, 'T.START': 73309488, 'T.END': 73309506, 'CG': '18='}\n",
+ "7046706 D134#1#chr03 3230 3276\n",
+ "\tIn path\n",
+ "\t 73309509 73309555\n",
+ "{'Q.START': 3229, 'Q.END': 3230, 'T.START': 73309507, 'T.END': 73309508, 'CG': '1='}\n",
+ "7046707 D134#1#chr03 3276 3277\n",
+ "\tNot in path\n",
+ "7046709 D134#1#chr03 3277 3315\n",
+ "\tIn path\n",
+ "\t 73309558 73309596\n",
+ "{'Q.START': 3230, 'Q.END': 3276, 'T.START': 73309509, 'T.END': 73309555, 'CG': '46='}\n",
+ "7046710 D134#1#chr03 3315 3316\n",
+ "\tNot in path\n",
+ "7046712 D134#1#chr03 3316 3322\n",
+ "\tIn path\n",
+ "\t 73309599 73309605\n",
+ "{'Q.START': 3277, 'Q.END': 3315, 'T.START': 73309558, 'T.END': 73309596, 'CG': '38='}\n",
+ "7046713 D134#1#chr03 3322 3323\n",
+ "\tNot in path\n",
+ "7046715 D134#1#chr03 3323 3348\n",
+ "\tIn path\n",
+ "\t 73309608 73309633\n",
+ "{'Q.START': 3316, 'Q.END': 3322, 'T.START': 73309599, 'T.END': 73309605, 'CG': '6='}\n",
+ "7046718 D134#1#chr03 3352 3353\n",
+ "\tIn path\n",
+ "\t 73309634 73309635\n",
+ "{'Q.START': 3323, 'Q.END': 3348, 'T.START': 73309608, 'T.END': 73309633, 'CG': '25='}\n",
+ "7046717 D134#1#chr03 3351 3352\n",
+ "\tIn path\n",
+ "\t 73309636 73309637\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
+ "7046718 D134#1#chr03 3352 3353\n",
+ "\tIn path\n",
+ "\t 73309634 73309635\n",
+ "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}\n",
+ "7046717 D134#1#chr03 3351 3352\n",
+ "\tIn path\n",
+ "\t 73309636 73309637\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
+ "7046718 D134#1#chr03 3352 3353\n",
+ "\tIn path\n",
+ "\t 73309634 73309635\n",
+ "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}\n",
+ "7046720 D134#1#chr03 3353 3354\n",
+ "\tIn path\n",
+ "\t 73309638 73309639\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
+ "7046722 D134#1#chr03 3354 3356\n",
+ "\tIn path\n",
+ "\t 73309640 73309642\n",
+ "{'Q.START': 3353, 'Q.END': 3354, 'T.START': 73309638, 'T.END': 73309639, 'CG': '1='}\n",
+ "7046724 D134#1#chr03 3356 3357\n",
+ "\tNot in path\n",
+ "7046725 D134#1#chr03 3357 3489\n",
+ "\tIn path\n",
+ "\t 73309645 73309777\n",
+ "{'Q.START': 3354, 'Q.END': 3356, 'T.START': 73309640, 'T.END': 73309642, 'CG': '2='}\n",
+ "7046727 D134#1#chr03 3489 3490\n",
+ "\tNot in path\n",
+ "7046728 D134#1#chr03 3490 3642\n",
+ "\tIn path\n",
+ "\t 73309780 73309932\n",
+ "{'Q.START': 3357, 'Q.END': 3489, 'T.START': 73309645, 'T.END': 73309777, 'CG': '132='}\n",
+ "7046729 D134#1#chr03 3642 3644\n",
+ "\tNot in path\n",
+ "7046730 D134#1#chr03 3644 3685\n",
+ "\tIn path\n",
+ "\t 73309933 73309974\n",
+ "{'Q.START': 3490, 'Q.END': 3642, 'T.START': 73309780, 'T.END': 73309932, 'CG': '152='}\n",
+ "7046731 D134#1#chr03 3685 3687\n",
+ "\tNot in path\n",
+ "7046733 D134#1#chr03 3687 3693\n",
+ "\tIn path\n",
+ "\t 73309977 73309983\n",
+ "{'Q.START': 3644, 'Q.END': 3685, 'T.START': 73309933, 'T.END': 73309974, 'CG': '41='}\n",
+ "7046735 D134#1#chr03 3693 3694\n",
+ "\tNot in path\n",
+ "7046736 D134#1#chr03 3694 3708\n",
+ "\tIn path\n",
+ "\t 73309986 73310000\n",
+ "{'Q.START': 3687, 'Q.END': 3693, 'T.START': 73309977, 'T.END': 73309983, 'CG': '6='}\n",
+ "7046738 D134#1#chr03 3720 3721\n",
+ "\tIn path\n",
+ "\t 73310010 73310011\n",
+ "{'Q.START': 3694, 'Q.END': 3708, 'T.START': 73309986, 'T.END': 73310000, 'CG': '14='}\n",
+ "7046739 D134#1#chr03 3721 3722\n",
+ "\tIn path\n",
+ "\t 73310003 73310004\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
+ "7046740 D134#1#chr03 3716 3720\n",
+ "\tIn path\n",
+ "\t 73310005 73310009\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
+ "7046738 D134#1#chr03 3720 3721\n",
+ "\tIn path\n",
+ "\t 73310010 73310011\n",
+ "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}\n",
+ "7046739 D134#1#chr03 3721 3722\n",
+ "\tIn path\n",
+ "\t 73310003 73310004\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
+ "7046740 D134#1#chr03 3716 3720\n",
+ "\tIn path\n",
+ "\t 73310005 73310009\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
+ "7046738 D134#1#chr03 3720 3721\n",
+ "\tIn path\n",
+ "\t 73310010 73310011\n",
+ "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}\n",
+ "7046739 D134#1#chr03 3721 3722\n",
+ "\tIn path\n",
+ "\t 73310003 73310004\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
+ "7046741 D134#1#chr03 3722 3735\n",
+ "\tIn path\n",
+ "\t 73310012 73310045\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
+ "ALN_2\n",
+ "7594382 D134#1#chr03 0 1\n",
+ "\tIn path\n",
+ "\t 70220037 70220038\n",
+ "skipped\n",
+ "\n",
+ "7594369 D134#1#chr03 32 33\n",
+ "\tIn path\n",
+ "\t 70219216 70219217\n",
+ "{'Q.START': 0, 'Q.END': 1, 'T.START': 70220037, 'T.END': 70220038, 'CG': '1='}\n",
+ "7594371 D134#1#chr03 15 16\n",
+ "\tIn path\n",
+ "\t 70221163 70221164\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 15, 'Q.END': 16, 'T.START': 70221163, 'T.END': 70221164, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594356 D134#1#chr03 66 67\n",
+ "\tIn path\n",
+ "\t 70219570 70219571\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594626 D134#1#chr03 10 11\n",
+ "\tIn path\n",
+ "\t 70219214 70219215\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594011 D134#1#chr03 11 12\n",
+ "\tIn path\n",
+ "\t 70219995 70219996\n",
+ "{'Q.START': 10, 'Q.END': 11, 'T.START': 70219214, 'T.END': 70219215, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 11, 'Q.END': 12, 'T.START': 70219995, 'T.END': 70219996, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594369 D134#1#chr03 32 33\n",
+ "\tIn path\n",
+ "\t 70219216 70219217\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594371 D134#1#chr03 15 16\n",
+ "\tIn path\n",
+ "\t 70221163 70221164\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 15, 'Q.END': 16, 'T.START': 70221163, 'T.END': 70221164, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594241 D134#1#chr03 20 21\n",
+ "\tIn path\n",
+ "\t 70219220 70219221\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594248 D134#1#chr03 21 22\n",
+ "\tNot in path\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 20, 'Q.END': 21, 'T.START': 70219220, 'T.END': 70219221, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594315 D134#1#chr03 53 54\n",
+ "\tIn path\n",
+ "\t 70219857 70219858\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
+ "7594330 D134#1#chr03 26 27\n",
+ "\tNot in path\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594315 D134#1#chr03 53 54\n",
+ "\tIn path\n",
+ "\t 70219857 70219858\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594369 D134#1#chr03 32 33\n",
+ "\tIn path\n",
+ "\t 70219216 70219217\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
+ "7594026 D134#1#chr03 37 38\n",
+ "\tIn path\n",
+ "\t 70220249 70220250\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 37, 'Q.END': 38, 'T.START': 70220249, 'T.END': 70220250, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594026 D134#1#chr03 37 38\n",
+ "\tIn path\n",
+ "\t 70220249 70220250\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 37, 'Q.END': 38, 'T.START': 70220249, 'T.END': 70220250, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594315 D134#1#chr03 53 54\n",
+ "\tIn path\n",
+ "\t 70219857 70219858\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594356 D134#1#chr03 66 67\n",
+ "\tIn path\n",
+ "\t 70219570 70219571\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594356 D134#1#chr03 66 67\n",
+ "\tIn path\n",
+ "\t 70219570 70219571\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594350 D134#1#chr03 70 71\n",
+ "\tIn path\n",
+ "\t 70219226 70219227\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594264 D134#1#chr03 71 72\n",
+ "\tIn path\n",
+ "\t 70219228 70219229\n",
+ "{'Q.START': 70, 'Q.END': 71, 'T.START': 70219226, 'T.END': 70219227, 'CG': '1='}\n",
+ "7594207 D134#1#chr03 72 73\n",
+ "\tIn path\n",
+ "\t 70219230 70219231\n",
+ "{'Q.START': 71, 'Q.END': 72, 'T.START': 70219228, 'T.END': 70219229, 'CG': '1='}\n",
+ "7594225 D134#1#chr03 73 74\n",
+ "\tIn path\n",
+ "\t 70219232 70219233\n",
+ "{'Q.START': 72, 'Q.END': 73, 'T.START': 70219230, 'T.END': 70219231, 'CG': '1='}\n",
+ "7594227 D134#1#chr03 74 75\n",
+ "\tIn path\n",
+ "\t 70220150 70220151\n",
+ "{'Q.START': 73, 'Q.END': 74, 'T.START': 70219232, 'T.END': 70219233, 'CG': '1='}\n",
+ "7594120 D134#1#chr03 75 76\n",
+ "\tIn path\n",
+ "\t 70219236 70219237\n",
+ "{'Q.START': 74, 'Q.END': 75, 'T.START': 70220150, 'T.END': 70220151, 'CG': '1='}\n",
+ "7594132 D134#1#chr03 76 77\n",
+ "\tIn path\n",
+ "\t 70219777 70219778\n",
+ "{'Q.START': 75, 'Q.END': 76, 'T.START': 70219236, 'T.END': 70219237, 'CG': '1='}\n",
+ "7594165 D134#1#chr03 77 78\n",
+ "\tIn path\n",
+ "\t 70219240 70219241\n",
+ "{'Q.START': 76, 'Q.END': 77, 'T.START': 70219777, 'T.END': 70219778, 'CG': '1='}\n",
+ "7594172 D134#1#chr03 78 3735\n",
+ "\tNot in path\n"
+ ]
+ }
+ ],
+ "source": [
+ "ALNS = {}\n",
+ "## Iterating over alignments\n",
+ "for aln_name in aln_dict.keys():\n",
+ " \n",
+ " ## Iterating over paths of the gfa\n",
+ " for path_name in paths.keys():\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(aln_name)\n",
+ " _ = [] # Temporary list holding alignment blocks\n",
+ "\n",
+ " ## Iterating over alignment nodes of the current alignment\n",
+ " for node_id, orient in aln_dict[aln_name][\"PATH.MATCH\"]:\n",
+ "\n",
+ " # Getting node info\n",
+ " n_info = nodes[node_id]\n",
+ " q_start = n_info[aln_name][\"START\"] # Start position on the query\n",
+ " q_end = n_info[aln_name][\"END\"] # End position on the query\n",
+ " _CG = n_info[aln_name][\"CIGAR\"] # Cigar of the alignment on the current node\n",
+ "\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(node_id, path_name, q_start, q_end)\n",
+ "\n",
+ " ## Checking if path is traversing the current node\n",
+ " if path_name in list(n_info.keys()):\n",
+ " if path_name == \"D134#1#chr03\": print(\"\\tIn path\")\n",
+ "\n",
+ " ## Getting start and end position on the target given the orientation of the node in the alignment and the path\n",
+ " if n_info[aln_name][\"STRAND\"] == n_info[path_name][\"STRAND\"] :\n",
+ " t_start = n_info[path_name][\"START\"]+n_info[aln_name][\"S.OFF\"]\n",
+ " t_end = n_info[path_name][\"END\"]+n_info[aln_name][\"E.OFF\"] \n",
+ " else :\n",
+ " t_end = n_info[path_name][\"START\"]+n_info[aln_name][\"S.OFF\"]\n",
+ " t_start = n_info[path_name][\"END\"]+n_info[aln_name][\"E.OFF\"]\n",
+ "\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"\\t\", t_start, t_end)\n",
+ "\n",
+ " \"\"\"\n",
+ " If the latest block t.end and q.end matches with the current node t.start and q.start, \n",
+ " the node should be added to the block. Else, we terminate the block and add the node to a new block\n",
+ " \"\"\"\n",
+ " \n",
+ " # Non empty temporary list of aln and ending of the last block is the same as the start of the new node : \n",
+ " if len(_) and _[-1][\"T.END\"] == t_start and _[-1][\"Q.END\"]+1 == q_start: \n",
+ " tmp_aln[\"Q.END\"] = q_end\n",
+ " tmp_aln[\"T.END\"] = t_end\n",
+ " tmp_aln[\"CG\"] += _CG\n",
+ "# elif len(_) and _[-1][\"T.END\"] == t_start: # Following on the target not on the query (i.e. Insertion)\n",
+ "# tmp_aln[\"T.END\"] = t_end\n",
+ "# tmp_aln[\"CG\"] += f\"{nodes_length[node_id]}I\"\n",
+ "# elif len(_) and _[-1][\"Q.END\"]+1 == q_start: # Following on the query, not on the target (i.e. Deletion)\n",
+ "# tmp_aln[\"Q.END\"] = q_end\n",
+ "# tmp_aln[\"CG\"] += f\"{nodes_length[node_id]}D\"\n",
+ " else : # Else, completely different\n",
+ " try : \n",
+ " _.append(tmp_aln)\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(tmp_aln)\n",
+ " except : \n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"skipped\\n\")\n",
+ " tmp_aln = {\n",
+ " \"Q.START\": q_start,\n",
+ " \"Q.END\": q_end,\n",
+ " \"T.START\": t_start,\n",
+ " \"T.END\": t_end,\n",
+ " \"CG\": _CG,\n",
+ " }\n",
+ " \n",
+ " else : \n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"\\tNot in path\")\n",
+ " # Node is not in the path\n",
+ "\n",
+ " del tmp_aln\n",
+ " \n",
+ " ALNS[(path_name, aln_name)] = _"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 9,
+ "id": "547f03fa-cbd5-42f9-b668-1ca4404795ba",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[{'Q.START': 0, 'Q.END': 77, 'T.START': 73306158, 'T.END': 73306235, 'CG': '77='}, {'Q.START': 77, 'Q.END': 82, 'T.START': 73306238, 'T.END': 73306243, 'CG': '5='}, {'Q.START': 83, 'Q.END': 138, 'T.START': 73306246, 'T.END': 73306301, 'CG': '55='}, {'Q.START': 139, 'Q.END': 202, 'T.START': 73306302, 'T.END': 73306365, 'CG': '63='}, {'Q.START': 202, 'Q.END': 203, 'T.START': 73306366, 'T.END': 73306367, 'CG': '1='}, {'Q.START': 203, 'Q.END': 379, 'T.START': 73306368, 'T.END': 73306544, 'CG': '176='}, {'Q.START': 379, 'Q.END': 380, 'T.START': 73306545, 'T.END': 73306546, 'CG': '1='}, {'Q.START': 380, 'Q.END': 429, 'T.START': 73306547, 'T.END': 73306596, 'CG': '49='}, {'Q.START': 429, 'Q.END': 430, 'T.START': 73306597, 'T.END': 73306598, 'CG': '1='}, {'Q.START': 430, 'Q.END': 457, 'T.START': 73306599, 'T.END': 73306626, 'CG': '27='}, {'Q.START': 457, 'Q.END': 492, 'T.START': 73306641, 'T.END': 73306676, 'CG': '35='}, {'Q.START': 508, 'Q.END': 564, 'T.START': 73306694, 'T.END': 73306750, 'CG': '56='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 569, 'Q.END': 824, 'T.START': 73306755, 'T.END': 73307010, 'CG': '255='}, {'Q.START': 826, 'Q.END': 858, 'T.START': 73307011, 'T.END': 73307043, 'CG': '32='}, {'Q.START': 858, 'Q.END': 859, 'T.START': 73307044, 'T.END': 73307045, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 869, 'Q.END': 913, 'T.START': 73307048, 'T.END': 73307092, 'CG': '44='}, {'Q.START': 913, 'Q.END': 919, 'T.START': 73307093, 'T.END': 73307099, 'CG': '6='}, {'Q.START': 919, 'Q.END': 978, 'T.START': 73307100, 'T.END': 73307159, 'CG': '59='}, {'Q.START': 978, 'Q.END': 979, 'T.START': 73307160, 'T.END': 73307161, 'CG': '1='}, {'Q.START': 979, 'Q.END': 1038, 'T.START': 73307162, 'T.END': 73307221, 'CG': '59='}, {'Q.START': 1038, 'Q.END': 1045, 'T.START': 73307224, 'T.END': 73307231, 'CG': '7='}, {'Q.START': 1045, 'Q.END': 1046, 'T.START': 73307232, 'T.END': 73307233, 'CG': '1='}, {'Q.START': 1046, 'Q.END': 1080, 'T.START': 73307234, 'T.END': 73307268, 'CG': '34='}, {'Q.START': 1080, 'Q.END': 1081, 'T.START': 73307269, 'T.END': 73307270, 'CG': '1='}, {'Q.START': 1081, 'Q.END': 1107, 'T.START': 73307271, 'T.END': 73307297, 'CG': '26='}, {'Q.START': 1108, 'Q.END': 1183, 'T.START': 73307300, 'T.END': 73307375, 'CG': '75='}, {'Q.START': 1183, 'Q.END': 1186, 'T.START': 73307376, 'T.END': 73307379, 'CG': '3='}, {'Q.START': 1224, 'Q.END': 1257, 'T.START': 73307419, 'T.END': 73307452, 'CG': '33='}, {'Q.START': 1289, 'Q.END': 1311, 'T.START': 73307475, 'T.END': 73307497, 'CG': '22='}, {'Q.START': 1359, 'Q.END': 1382, 'T.START': 73307546, 'T.END': 73307569, 'CG': '23='}, {'Q.START': 1434, 'Q.END': 1451, 'T.START': 73307643, 'T.END': 73307660, 'CG': '17='}, {'Q.START': 1451, 'Q.END': 1531, 'T.START': 73307661, 'T.END': 73307741, 'CG': '80='}, {'Q.START': 1532, 'Q.END': 1543, 'T.START': 73307744, 'T.END': 73307755, 'CG': '11='}, {'Q.START': 1544, 'Q.END': 1572, 'T.START': 73307758, 'T.END': 73307786, 'CG': '28='}, {'Q.START': 1572, 'Q.END': 1573, 'T.START': 73307787, 'T.END': 73307788, 'CG': '1='}, {'Q.START': 1573, 'Q.END': 1587, 'T.START': 73307789, 'T.END': 73307803, 'CG': '14='}, {'Q.START': 1588, 'Q.END': 1616, 'T.START': 73307806, 'T.END': 73307834, 'CG': '28='}, {'Q.START': 1616, 'Q.END': 1617, 'T.START': 73307835, 'T.END': 73307836, 'CG': '1='}, {'Q.START': 1617, 'Q.END': 1646, 'T.START': 73307837, 'T.END': 73307866, 'CG': '29='}, {'Q.START': 1646, 'Q.END': 1661, 'T.START': 73307867, 'T.END': 73307882, 'CG': '15='}, {'Q.START': 1661, 'Q.END': 1673, 'T.START': 73307883, 'T.END': 73307895, 'CG': '12='}, {'Q.START': 1673, 'Q.END': 1674, 'T.START': 73307896, 'T.END': 73307897, 'CG': '1='}, {'Q.START': 1674, 'Q.END': 1726, 'T.START': 73307898, 'T.END': 73307950, 'CG': '52='}, {'Q.START': 1727, 'Q.END': 1762, 'T.START': 73307953, 'T.END': 73307988, 'CG': '35='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1767, 'Q.END': 1824, 'T.START': 73307995, 'T.END': 73308052, 'CG': '57='}, {'Q.START': 1824, 'Q.END': 1825, 'T.START': 73308053, 'T.END': 73308054, 'CG': '1='}, {'Q.START': 1825, 'Q.END': 1975, 'T.START': 73308055, 'T.END': 73308205, 'CG': '150='}, {'Q.START': 1976, 'Q.END': 2015, 'T.START': 73308208, 'T.END': 73308247, 'CG': '39='}, {'Q.START': 2016, 'Q.END': 2047, 'T.START': 73308250, 'T.END': 73308281, 'CG': '31='}, {'Q.START': 2047, 'Q.END': 2055, 'T.START': 73308286, 'T.END': 73308294, 'CG': '8='}, {'Q.START': 2056, 'Q.END': 2120, 'T.START': 73308297, 'T.END': 73308361, 'CG': '64='}, {'Q.START': 2120, 'Q.END': 2121, 'T.START': 73308362, 'T.END': 73308363, 'CG': '1='}, {'Q.START': 2121, 'Q.END': 2157, 'T.START': 73308364, 'T.END': 73308400, 'CG': '36='}, {'Q.START': 2158, 'Q.END': 2170, 'T.START': 73308403, 'T.END': 73308415, 'CG': '12='}, {'Q.START': 2170, 'Q.END': 2171, 'T.START': 73308416, 'T.END': 73308417, 'CG': '1='}, {'Q.START': 2171, 'Q.END': 2205, 'T.START': 73308418, 'T.END': 73308452, 'CG': '34='}, {'Q.START': 2206, 'Q.END': 2344, 'T.START': 73308455, 'T.END': 73308593, 'CG': '138='}, {'Q.START': 2345, 'Q.END': 2364, 'T.START': 73308596, 'T.END': 73308615, 'CG': '19='}, {'Q.START': 2364, 'Q.END': 2383, 'T.START': 73308616, 'T.END': 73308635, 'CG': '19='}, {'Q.START': 2383, 'Q.END': 2408, 'T.START': 73308636, 'T.END': 73308661, 'CG': '25='}, {'Q.START': 2408, 'Q.END': 2409, 'T.START': 73308662, 'T.END': 73308663, 'CG': '1='}, {'Q.START': 2409, 'Q.END': 2441, 'T.START': 73308664, 'T.END': 73308696, 'CG': '32='}, {'Q.START': 2441, 'Q.END': 2442, 'T.START': 73308697, 'T.END': 73308698, 'CG': '1='}, {'Q.START': 2442, 'Q.END': 2580, 'T.START': 73308699, 'T.END': 73308837, 'CG': '138='}, {'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}, {'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}, {'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}, {'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}, {'Q.START': 2584, 'Q.END': 2764, 'T.START': 73308842, 'T.END': 73309022, 'CG': '180='}, {'Q.START': 2765, 'Q.END': 2797, 'T.START': 73309025, 'T.END': 73309057, 'CG': '32='}, {'Q.START': 2798, 'Q.END': 2878, 'T.START': 73309060, 'T.END': 73309140, 'CG': '80='}, {'Q.START': 2878, 'Q.END': 2879, 'T.START': 73309141, 'T.END': 73309142, 'CG': '1='}, {'Q.START': 2879, 'Q.END': 2951, 'T.START': 73309143, 'T.END': 73309215, 'CG': '72='}, {'Q.START': 2951, 'Q.END': 2952, 'T.START': 73309216, 'T.END': 73309217, 'CG': '1='}, {'Q.START': 2952, 'Q.END': 3002, 'T.START': 73309218, 'T.END': 73309268, 'CG': '50='}, {'Q.START': 3002, 'Q.END': 3077, 'T.START': 73309271, 'T.END': 73309346, 'CG': '75='}, {'Q.START': 3077, 'Q.END': 3078, 'T.START': 73309347, 'T.END': 73309348, 'CG': '1='}, {'Q.START': 3078, 'Q.END': 3093, 'T.START': 73309349, 'T.END': 73309364, 'CG': '15='}, {'Q.START': 3094, 'Q.END': 3097, 'T.START': 73309367, 'T.END': 73309370, 'CG': '3='}, {'Q.START': 3097, 'Q.END': 3140, 'T.START': 73309371, 'T.END': 73309414, 'CG': '43='}, {'Q.START': 3140, 'Q.END': 3210, 'T.START': 73309415, 'T.END': 73309485, 'CG': '70='}, {'Q.START': 3210, 'Q.END': 3211, 'T.START': 73309486, 'T.END': 73309487, 'CG': '1='}, {'Q.START': 3211, 'Q.END': 3229, 'T.START': 73309488, 'T.END': 73309506, 'CG': '18='}, {'Q.START': 3229, 'Q.END': 3230, 'T.START': 73309507, 'T.END': 73309508, 'CG': '1='}, {'Q.START': 3230, 'Q.END': 3276, 'T.START': 73309509, 'T.END': 73309555, 'CG': '46='}, {'Q.START': 3277, 'Q.END': 3315, 'T.START': 73309558, 'T.END': 73309596, 'CG': '38='}, {'Q.START': 3316, 'Q.END': 3322, 'T.START': 73309599, 'T.END': 73309605, 'CG': '6='}, {'Q.START': 3323, 'Q.END': 3348, 'T.START': 73309608, 'T.END': 73309633, 'CG': '25='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3353, 'Q.END': 3354, 'T.START': 73309638, 'T.END': 73309639, 'CG': '1='}, {'Q.START': 3354, 'Q.END': 3356, 'T.START': 73309640, 'T.END': 73309642, 'CG': '2='}, {'Q.START': 3357, 'Q.END': 3489, 'T.START': 73309645, 'T.END': 73309777, 'CG': '132='}, {'Q.START': 3490, 'Q.END': 3642, 'T.START': 73309780, 'T.END': 73309932, 'CG': '152='}, {'Q.START': 3644, 'Q.END': 3685, 'T.START': 73309933, 'T.END': 73309974, 'CG': '41='}, {'Q.START': 3687, 'Q.END': 3693, 'T.START': 73309977, 'T.END': 73309983, 'CG': '6='}, {'Q.START': 3694, 'Q.END': 3708, 'T.START': 73309986, 'T.END': 73310000, 'CG': '14='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}, {'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}, {'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}]\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(ALNS[(\"D134#1#chr03\", \"ALN_1\")])"
+ ]
+ }
+ ],
+ "metadata": {
+ "kernelspec": {
+ "display_name": "Python 3 (ipykernel)",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3",
+ "version": "3.10.14"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/gaf2aln.ipynb b/gaf2aln.ipynb
new file mode 100644
index 0000000..04fe866
--- /dev/null
+++ b/gaf2aln.ipynb
@@ -0,0 +1,2443 @@
+{
+ "cells": [
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "id": "4ffaf9f6-cc1e-4190-9351-5431c930d25b",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import numpy as np\n",
+ "import pandas as pd\n",
+ "import argparse\n",
+ "import concurrent.futures\n",
+ "import os\n",
+ "import re\n",
+ "\n",
+ "# Replace for argparse arguments\n",
+ "class arguments():\n",
+ " gfa = \"/home/amergez/Documents/Scratch/LeChou/35Bra-v2a/35Bra-v2a.chr03.gfa\"\n",
+ " gaf = \"/home/amergez/Documents/Scratch/LeChou/35Bra-v2a/Mapping2Graph/GA.FLC2.aln.gaf\"\n",
+ " threads = 8\n",
+ " version = False\n",
+ "args = arguments()"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 2,
+ "id": "280c8847-22e8-4063-bde8-3e4e72cf20e7",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# Toolbox\n",
+ "def walk2path(walk):\n",
+ " \"\"\"\n",
+ " Takes a walk in a single string and returns a list of nodes id with signs (gfa v1 like)\n",
+ " \"\"\"\n",
+ " _ = re.findall(r'>\\w+|<\\w+', walk)\n",
+ " # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'\n",
+ " return [f'{elem[1:]}{(elem[0] == \">\")*\"+\"+(elem[0] == \"<\")*\"-\"}' for elem in _]\n",
+ "\n",
+ "def cigar2basealn(cigar):\n",
+ " \"\"\"\n",
+ " Takes a CIGAR string and convert it into a list of base level alignment.\n",
+ " For example : \"345=\" -> [\"=\", \"=\", ..., \"=\"] of length 345.\n",
+ " \"\"\"\n",
+ " _ = re.findall(r'\\d+\\D', cigar)\n",
+ " final_cigar = []\n",
+ " for match in _:\n",
+ " final_cigar += [match[-1]]*int(match[:-1])\n",
+ "\n",
+ " return final_cigar\n",
+ "\n",
+ "def basealn2cigar(base_aln_list):\n",
+ " \n",
+ " last_elem = base_aln_list[0]\n",
+ " CIGAR = [[1, last_elem]]\n",
+ " for elem in base_aln_list[1:]:\n",
+ " if elem == last_elem:\n",
+ " CIGAR[-1][0] += 1\n",
+ "\n",
+ " else :\n",
+ " CIGAR[-1][0] = str(CIGAR[-1][0])\n",
+ " CIGAR.append([1, elem])\n",
+ " last_elem = elem\n",
+ " CIGAR[-1][0] = str(CIGAR[-1][0])\n",
+ " return \"\".join([\"\".join(block) for block in CIGAR if block[1] != \"\"])"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 3,
+ "id": "be12e9d4-de76-4c8b-af84-6567549483f4",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::GAF Parser] Reading /home/amergez/Documents/Scratch/LeChou/35Bra-v2a/Mapping2Graph/GA.FLC2.aln.gaf ...\n",
+ "[gaf2aln::GAF Parser] Extracting alignments ...\n",
+ "{'ALN_1': {'QRY.NAME': 'FLC2.TO1000#1#chr03', 'QRY.LEN': '3735', 'QRY.START': '0', 'QRY.END': '3735', 'STRAND': '+', 'PATH.MATCH': [('7046526', '+'), ('7046528', '+'), ('7046530', '+'), ('7046531', '+'), ('7046532', '+'), ('7046533', '+'), ('7046534', '+'), ('7046536', '+'), ('7046537', '+'), ('7046539', '+'), ('7046541', '+'), ('7046542', '+'), ('7046544', '+'), ('7046546', '+'), ('7046547', '+'), ('7046549', '+'), ('7046551', '+'), ('7046552', '+'), ('7046554', '+'), ('7046556', '+'), ('7046556', '+'), ('7046556', '+'), ('7046557', '+'), ('7046558', '+'), ('7046559', '+'), ('7046560', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046562', '+'), ('7046564', '+'), ('7046565', '+'), ('7046567', '+'), ('7046568', '+'), ('7046570', '+'), ('7046571', '+'), ('7046573', '+'), ('7046574', '+'), ('7046576', '+'), ('7046577', '+'), ('7046579', '+'), ('7046581', '+'), ('7046583', '+'), ('7046584', '+'), ('7046586', '+'), ('7046587', '+'), ('7046589', '+'), ('7046590', '+'), ('7046592', '+'), ('7046593', '+'), ('7046594', '+'), ('7046596', '+'), ('7046597', '+'), ('7046599', '+'), ('7046600', '+'), ('7046601', '+'), ('7046603', '+'), ('7046604', '+'), ('7046606', '+'), ('7046608', '+'), ('7046609', '+'), ('7046621', '+'), ('7046622', '+'), ('7046624', '+'), ('7046625', '+'), ('7046626', '+'), ('7046628', '+'), ('7046631', '+'), ('7046673', '+'), ('7046631', '+'), ('7046673', '+'), ('7046631', '+'), ('7046632', '+'), ('7046634', '+'), ('7046635', '+'), ('7046637', '+'), ('7046638', '+'), ('7046639', '+'), ('7046641', '+'), ('7046644', '+'), ('7046646', '+'), ('7046647', '+'), ('7046649', '+'), ('7046650', '+'), ('7046652', '+'), ('7046653', '+'), ('7046654', '+'), ('7046656', '+'), ('7046657', '+'), ('7046659', '+'), ('7046660', '+'), ('7046662', '+'), ('7046663', '+'), ('7046665', '+'), ('7046667', '+'), ('7046668', '+'), ('7046670', '+'), ('7046671', '+'), ('7046674', '+'), ('7046675', '+'), ('7046674', '+'), ('7046675', '+'), ('7046676', '+'), ('7046678', '+'), ('7046679', '+'), ('7046680', '+'), ('7046682', '+'), ('7046684', '+'), ('7046685', '+'), ('7046686', '+'), ('7046688', '+'), ('7046690', '+'), ('7046692', '+'), ('7046693', '+'), ('7046695', '+'), ('7046696', '+'), ('7046698', '+'), ('7046700', '+'), ('7046702', '+'), ('7046703', '+'), ('7046704', '+'), ('7046706', '+'), ('7046707', '+'), ('7046709', '+'), ('7046710', '+'), ('7046712', '+'), ('7046713', '+'), ('7046715', '+'), ('7046718', '+'), ('7046717', '+'), ('7046718', '+'), ('7046717', '+'), ('7046718', '+'), ('7046720', '+'), ('7046722', '+'), ('7046724', '+'), ('7046725', '+'), ('7046727', '+'), ('7046728', '+'), ('7046729', '+'), ('7046730', '+'), ('7046731', '+'), ('7046733', '+'), ('7046735', '+'), ('7046736', '+'), ('7046738', '+'), ('7046739', '+'), ('7046740', '+'), ('7046738', '+'), ('7046739', '+'), ('7046740', '+'), ('7046738', '+'), ('7046739', '+'), ('7046741', '+')], 'PATH.LEN': '3822', 'ALN.START': '77', 'ALN.END': '3812', 'RES.MATCH': '3735', 'ALN.BLOCK.LEN': '3735', 'MAPPING.QUAL': '60', 'RAW.CIGAR': 'cg:Z:3735=', 'TAGS': 'AS:f:3735,dv:f:0,id:f:1'}, 'ALN_2': {'QRY.NAME': 'FLC2.TO1000#1#chr03', 'QRY.LEN': '3735', 'QRY.START': '0', 'QRY.END': '3735', 'STRAND': '+', 'PATH.MATCH': [('7594382', '+'), ('7594369', '+'), ('7594371', '+'), ('7594021', '+'), ('7594286', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594374', '+'), ('7594375', '+'), ('7594626', '+'), ('7594011', '+'), ('7594374', '+'), ('7594375', '+'), ('7594369', '+'), ('7594371', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594241', '+'), ('7594248', '+'), ('7594286', '+'), ('7594311', '+'), ('7594315', '+'), ('7594311', '+'), ('7594330', '+'), ('7594311', '+'), ('7594315', '+'), ('7594374', '+'), ('7594311', '+'), ('7594374', '+'), ('7594369', '+'), ('7594021', '+'), ('7594026', '+'), ('7594021', '+'), ('7594021', '+'), ('7594026', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594286', '+'), ('7594374', '+'), ('7594021', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594315', '+'), ('7594286', '+'), ('7594311', '+'), ('7594374', '+'), ('7594021', '+'), ('7594286', '+'), ('7594286', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594374', '+'), ('7594375', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594375', '+'), ('7594374', '+'), ('7594350', '+'), ('7594264', '+'), ('7594207', '+'), ('7594225', '+'), ('7594227', '+'), ('7594120', '+'), ('7594132', '+'), ('7594165', '+'), ('7594172', '+')], 'PATH.LEN': '61224', 'ALN.START': '0', 'ALN.END': '3735', 'RES.MATCH': '3734', 'ALN.BLOCK.LEN': '3735', 'MAPPING.QUAL': '0', 'RAW.CIGAR': 'cg:Z:57=1X3677=', 'TAGS': 'AS:f:3732.06,dv:f:0.000267738,id:f:0.999732'}}\n"
+ ]
+ }
+ ],
+ "source": [
+ "# Parsing the .gaf file\n",
+ "print(f\"[gaf2aln::GAF Parser] Reading {args.gaf} ...\")\n",
+ "with open(args.gaf, 'r') as file:\n",
+ " gaf_lines = file.readlines()\n",
+ "\n",
+ "gaf_col = [\n",
+ " \"QRY.NAME\", \"QRY.LEN\", \"QRY.START\", \"QRY.END\", \"STRAND\", \n",
+ " \"PATH.MATCH\", \"PATH.LEN\", \"ALN.START\", \"ALN.END\",\n",
+ " \"RES.MATCH\", \"ALN.BLOCK.LEN\", \"MAPPING.QUAL\"\n",
+ " ]\n",
+ "\n",
+ "# Creating dictionnary to store alignments\n",
+ "print(f\"[gaf2aln::GAF Parser] Extracting alignments ...\")\n",
+ "aln_dict = {}\n",
+ "for line in range(len(gaf_lines)):\n",
+ " ## Splitting the line by tabulation\n",
+ " line_content = gaf_lines[line][:-1].split('\\t')\n",
+ "\n",
+ " ## Adding alignement info to dictionnary\n",
+ " aln_dict[f\"ALN_{line+1}\"] = {\n",
+ " gaf_col[i]: line_content[i] for i in range(len(gaf_col))\n",
+ " }\n",
+ " \n",
+ " ## Splitting \"PATH.MATCH\" into a list\n",
+ " aln_dict[f\"ALN_{line+1}\"][\"PATH.MATCH\"] = [\n",
+ " (str(node_id[:-1]), node_id[-1]) \n",
+ " for node_id in walk2path(aln_dict[f\"ALN_{line+1}\"][\"PATH.MATCH\"])\n",
+ " ]\n",
+ "\n",
+ " ## Adding CIGAR\n",
+ " aln_dict[f\"ALN_{line+1}\"][\"RAW.CIGAR\"] = line_content[-1]\n",
+ "\n",
+ " ## Adding tags\n",
+ " aln_dict[f\"ALN_{line+1}\"][\"TAGS\"] = \",\".join(line_content[13:-1])\n",
+ "\n",
+ "# Getting nodes of interest ids\n",
+ "aln_nodes = np.unique([\n",
+ " str(node_id) \n",
+ " for aln in aln_dict.keys() \n",
+ " for node_id, orient in aln_dict[aln][\"PATH.MATCH\"]\n",
+ "]).tolist()\n",
+ "\n",
+ "print(aln_dict)\n",
+ "del gaf_lines, gaf_col"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 4,
+ "id": "2f891424-0d88-4fd3-99ff-b0a8c90587ff",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::GFA Parser] Reading /home/amergez/Documents/Scratch/LeChou/35Bra-v2a/35Bra-v2a.chr03.gfa ...\n",
+ "[gaf2aln::GFA Parser] Extracting nodes, paths and links ...\n"
+ ]
+ }
+ ],
+ "source": [
+ "# Parsing the .gfa\n",
+ "print(f\"[gaf2aln::GFA Parser] Reading {args.gfa} ...\")\n",
+ "with open(args.gfa, 'r') as file:\n",
+ " gfa_lines = file.readlines()\n",
+ "\n",
+ "# Nodes length dictionnary structured as follow :\n",
+ "# {<NODE.ID>: <NODE.LENGTH>}\n",
+ "nodes_length = {}\n",
+ "# Nodes dictionnary structured as follow :\n",
+ "# { <ALN.NODE.ID> : {\n",
+ "# <PATH.NAME>: {\"START\": start, \"END\": end, \"STRAND\": strand), \n",
+ "# <ALN.NAME>: {\"START\": start, \"END\": end, \"S.OFF\": start.offset, \"E.OFF\": end.offset, \"STRAND\": strand, \"CIGAR\": CIGAR}\n",
+ "# }\n",
+ "# }\n",
+ "nodes = {node_id: {} for node_id in aln_nodes}\n",
+ "# Paths dictionnary structured as follow :\n",
+ "# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}\n",
+ "paths = {}\n",
+ "# Links dictionnary structured as follow : \n",
+ "# {<FROM.NODE.ID>: {<TO.NODE.ID>: {FROM.ORIENT: <FROM.ORIENT>, TO.ORIENT: <TO.ORIENT>}}}\n",
+ "links = {}\n",
+ "\n",
+ "# Parsing the gfa\n",
+ "print(f\"[gaf2aln::GFA Parser] Extracting nodes, paths and links ...\")\n",
+ "\n",
+ "def GFA_parser(gfa_lines, nodes = nodes):\n",
+ " _links, _nodes, _nodes_length, paths = {}, {}, {}, {}\n",
+ " for line in gfa_lines:\n",
+ " line_content = line[:-1].split(\"\\t\")\n",
+ " line_id = line_content[0]\n",
+ " \n",
+ " # Segment line\n",
+ " if line_id == \"S\" :\n",
+ " \n",
+ " _nodes_length[str(line_content[1])] = len(line_content[2])\n",
+ " \n",
+ " # Link line\n",
+ " elif line_id == \"L\":\n",
+ " try :\n",
+ " _links[str(line_content[1])][str(line_content[3])] = {\n",
+ " \"FROM\": str(line_content[2]), \n",
+ " \"TO\": str(line_content[4])\n",
+ " }\n",
+ "\n",
+ " except :\n",
+ " _links[str(line_content[1])] = {\n",
+ " str(line_content[3]) : {\"FROM.ORIENT\": str(line_content[2]), \"TO.ORIENT\": str(line_content[4])}\n",
+ " }\n",
+ "\n",
+ " # Path line\n",
+ " elif line_id == \"P\":\n",
+ " _paths[str(line_content[1])] = {\n",
+ " \"NODES\": {\n",
+ " str(node_id[:-1]): str(node_id[-1])\n",
+ " for node_id in line_content[2].split(',')\n",
+ " },\n",
+ " \"CIGAR\": line_content[3]\n",
+ " }\n",
+ "\n",
+ " return [_links, _nodes, _nodes_length, _paths]\n",
+ "\n",
+ "# splits = np.quantile(range(len(gfa_lines)+1), q= np.array(args.threads+1)/args.threads, method='higher').tolist()\n",
+ "# res = []\n",
+ "# for i in range(1, len(splits)):\n",
+ "# res.append(executor.submit(GFA_parser, gfa_lines[splits[i-1]:splits[i]]))\n",
+ "\n",
+ "# for out in res:\n",
+ "# results = out.result()\n",
+ "\n",
+ "# for link_id, link_info in results[0].items():\n",
+ "# links[]\n",
+ "\n",
+ "\n",
+ "for line in gfa_lines:\n",
+ " line_content = line[:-1].split(\"\\t\")\n",
+ " line_id = line_content[0]\n",
+ " \n",
+ " # Segment line\n",
+ " if line_id == \"S\" :\n",
+ " \n",
+ " nodes_length[str(line_content[1])] = len(line_content[2])\n",
+ " \n",
+ " # Link line\n",
+ " elif line_id == \"L\":\n",
+ " try :\n",
+ " links[str(line_content[1])][str(line_content[3])] = {\n",
+ " \"FROM\": str(line_content[2]), \n",
+ " \"TO\": str(line_content[4])\n",
+ " }\n",
+ "\n",
+ " except :\n",
+ " links[str(line_content[1])] = {\n",
+ " str(line_content[3]) : {\"FROM.ORIENT\": str(line_content[2]), \"TO.ORIENT\": str(line_content[4])}\n",
+ " }\n",
+ "\n",
+ " # Path line\n",
+ " elif line_id == \"P\":\n",
+ " paths[str(line_content[1])] = {\n",
+ " \"NODES\": {\n",
+ " str(node_id[:-1]): str(node_id[-1])\n",
+ " for node_id in line_content[2].split(',')\n",
+ " },\n",
+ " \"CIGAR\": line_content[3]\n",
+ " }\n",
+ "\n",
+ "del gfa_lines"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 5,
+ "id": "a403c88e-54ea-4a67-9047-dc44eba7f51a",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::Graph position processing] Computing nodes positions in each paths...\n",
+ "[gaf2aln::Graph position processing] Running on Capitata#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on D101#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on D134#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on G06-09-28#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on G07-DH-33#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on HDEM#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on Korso#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on M249#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on OX-heart#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on PL021#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on RC34#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T02#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T03#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T04#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T06#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T07#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T08#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T09#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T10#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T11#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T12#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T13#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T14#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T15#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T16#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T17#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T18#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T19#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T21#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T24#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T25#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T26#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on T27#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on TO1000#1#chr03 ...\n",
+ "[gaf2aln::Graph position processing] Running on W1701#1#chr03 ...\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(f\"[gaf2aln::Graph position processing] Computing nodes positions in each paths...\")\n",
+ "def get_node_pos(path_name, nodes = nodes, paths = paths, nodes_length = nodes_length):\n",
+ " print(f\"[gaf2aln::Graph position processing] Running on {path_name} ...\")\n",
+ " cur_pos = 0\n",
+ "\n",
+ " out = {}\n",
+ " # Iterating over nodes in the path\n",
+ " for path_node in paths[path_name][\"NODES\"].keys():\n",
+ " # Instead of checking if the node is one interesting node, we try to add to the nodes dict\n",
+ " if path_node in aln_nodes :\n",
+ " out[path_node] = {\n",
+ " \"START\": cur_pos, # Start position of the node start in the currrent path\n",
+ " \"END\": cur_pos+nodes_length[path_node], # End position of the node end in the current path\n",
+ " \"STRAND\": paths[path_name][\"NODES\"][path_node] # Orientation of the node in the current path\n",
+ " } \n",
+ "\n",
+ " cur_pos += nodes_length[path_node]+1\n",
+ " else :\n",
+ " cur_pos += nodes_length[path_node]+1\n",
+ "\n",
+ " return out\n",
+ "\n",
+ "res = {}\n",
+ "executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)\n",
+ "# Adding nodes positions relative to path\n",
+ "for path_name in paths.keys():\n",
+ " res[path_name] = executor.submit(get_node_pos, path_name)\n",
+ "\n",
+ "executor.shutdown(wait=True)\n",
+ "\n",
+ "for path_name, out in res.items():\n",
+ " results = out.result()\n",
+ " for path_node, node_pos in results.items():\n",
+ " nodes[path_node][path_name] = node_pos\n",
+ "\n",
+ "del res"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 6,
+ "id": "bed36bd5-30eb-4d02-8b52-1ae5d753f8f8",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...\n",
+ "[gaf2aln::Alignment position processing] Running on ALN_1 ...\n",
+ "0 77 77 0 + 154 77\n",
+ "77 82 0 0 + 5 82\n",
+ "82 83 0 0 + 1 83\n",
+ "83 138 0 0 + 55 138\n",
+ "138 139 0 0 + 1 139\n",
+ "139 202 0 0 + 63 202\n",
+ "202 203 0 0 + 1 203\n",
+ "203 379 0 0 + 176 379\n",
+ "379 380 0 0 + 1 380\n",
+ "380 429 0 0 + 49 429\n",
+ "429 430 0 0 + 1 430\n",
+ "430 457 0 0 + 27 457\n",
+ "457 492 0 0 + 35 492\n",
+ "492 494 0 0 + 2 494\n",
+ "494 497 0 0 + 3 497\n",
+ "497 507 0 0 + 10 507\n",
+ "507 508 0 0 + 1 508\n",
+ "508 564 0 0 + 56 564\n",
+ "564 566 0 0 + 2 566\n",
+ "566 567 0 0 + 1 567\n",
+ "567 568 0 0 + 1 568\n",
+ "568 569 0 0 + 1 569\n",
+ "569 824 0 0 + 255 824\n",
+ "824 826 0 0 + 2 826\n",
+ "826 858 0 0 + 32 858\n",
+ "858 859 0 0 + 1 859\n",
+ "859 860 0 0 + 1 860\n",
+ "860 861 0 0 + 1 861\n",
+ "861 862 0 0 + 1 862\n",
+ "862 863 0 0 + 1 863\n",
+ "863 864 0 0 + 1 864\n",
+ "864 865 0 0 + 1 865\n",
+ "865 866 0 0 + 1 866\n",
+ "866 867 0 0 + 1 867\n",
+ "867 868 0 0 + 1 868\n",
+ "868 869 0 0 + 1 869\n",
+ "869 913 0 0 + 44 913\n",
+ "913 919 0 0 + 6 919\n",
+ "919 978 0 0 + 59 978\n",
+ "978 979 0 0 + 1 979\n",
+ "979 1038 0 0 + 59 1038\n",
+ "1038 1045 0 0 + 7 1045\n",
+ "1045 1046 0 0 + 1 1046\n",
+ "1046 1080 0 0 + 34 1080\n",
+ "1080 1081 0 0 + 1 1081\n",
+ "1081 1107 0 0 + 26 1107\n",
+ "1107 1108 0 0 + 1 1108\n",
+ "1108 1183 0 0 + 75 1183\n",
+ "1183 1186 0 0 + 3 1186\n",
+ "1186 1224 0 0 + 38 1224\n",
+ "1224 1257 0 0 + 33 1257\n",
+ "1257 1289 0 0 + 32 1289\n",
+ "1289 1311 0 0 + 22 1311\n",
+ "1311 1359 0 0 + 48 1359\n",
+ "1359 1382 0 0 + 23 1382\n",
+ "1382 1434 0 0 + 52 1434\n",
+ "1434 1451 0 0 + 17 1451\n",
+ "1451 1531 0 0 + 80 1531\n",
+ "1531 1532 0 0 + 1 1532\n",
+ "1532 1543 0 0 + 11 1543\n",
+ "1543 1544 0 0 + 1 1544\n",
+ "1544 1572 0 0 + 28 1572\n",
+ "1572 1573 0 0 + 1 1573\n",
+ "1573 1587 0 0 + 14 1587\n",
+ "1587 1588 0 0 + 1 1588\n",
+ "1588 1616 0 0 + 28 1616\n",
+ "1616 1617 0 0 + 1 1617\n",
+ "1617 1646 0 0 + 29 1646\n",
+ "1646 1661 0 0 + 15 1661\n",
+ "1661 1673 0 0 + 12 1673\n",
+ "1673 1674 0 0 + 1 1674\n",
+ "1674 1726 0 0 + 52 1726\n",
+ "1726 1727 0 0 + 1 1727\n",
+ "1727 1762 0 0 + 35 1762\n",
+ "1762 1763 0 0 + 1 1763\n",
+ "1763 1764 0 0 + 1 1764\n",
+ "1764 1765 0 0 + 1 1765\n",
+ "1765 1766 0 0 + 1 1766\n",
+ "1766 1767 0 0 + 1 1767\n",
+ "1767 1824 0 0 + 57 1824\n",
+ "1824 1825 0 0 + 1 1825\n",
+ "1825 1975 0 0 + 150 1975\n",
+ "1975 1976 0 0 + 1 1976\n",
+ "1976 2015 0 0 + 39 2015\n",
+ "2015 2016 0 0 + 1 2016\n",
+ "2016 2047 0 0 + 31 2047\n",
+ "2047 2055 0 0 + 8 2055\n",
+ "2055 2056 0 0 + 1 2056\n",
+ "2056 2120 0 0 + 64 2120\n",
+ "2120 2121 0 0 + 1 2121\n",
+ "2121 2157 0 0 + 36 2157\n",
+ "2157 2158 0 0 + 1 2158\n",
+ "2158 2170 0 0 + 12 2170\n",
+ "2170 2171 0 0 + 1 2171\n",
+ "2171 2205 0 0 + 34 2205\n",
+ "2205 2206 0 0 + 1 2206\n",
+ "2206 2344 0 0 + 138 2344\n",
+ "2344 2345 0 0 + 1 2345\n",
+ "2345 2364 0 0 + 19 2364\n",
+ "2364 2383 0 0 + 19 2383\n",
+ "2383 2408 0 0 + 25 2408\n",
+ "2408 2409 0 0 + 1 2409\n",
+ "2409 2441 0 0 + 32 2441\n",
+ "2441 2442 0 0 + 1 2442\n",
+ "2442 2580 0 0 + 138 2580\n",
+ "2580 2581 0 0 + 1 2581\n",
+ "2581 2582 0 0 + 1 2582\n",
+ "2582 2583 0 0 + 1 2583\n",
+ "2583 2584 0 0 + 1 2584\n",
+ "2584 2764 0 0 + 180 2764\n",
+ "2764 2765 0 0 + 1 2765\n",
+ "2765 2797 0 0 + 32 2797\n",
+ "2797 2798 0 0 + 1 2798\n",
+ "2798 2878 0 0 + 80 2878\n",
+ "2878 2879 0 0 + 1 2879\n",
+ "2879 2951 0 0 + 72 2951\n",
+ "2951 2952 0 0 + 1 2952\n",
+ "2952 3002 0 0 + 50 3002\n",
+ "3002 3077 0 0 + 75 3077\n",
+ "3077 3078 0 0 + 1 3078\n",
+ "3078 3093 0 0 + 15 3093\n",
+ "3093 3094 0 0 + 1 3094\n",
+ "3094 3097 0 0 + 3 3097\n",
+ "3097 3140 0 0 + 43 3140\n",
+ "3140 3210 0 0 + 70 3210\n",
+ "3210 3211 0 0 + 1 3211\n",
+ "3211 3229 0 0 + 18 3229\n",
+ "3229 3230 0 0 + 1 3230\n",
+ "3230 3276 0 0 + 46 3276\n",
+ "3276 3277 0 0 + 1 3277\n",
+ "3277 3315 0 0 + 38 3315\n",
+ "3315 3316 0 0 + 1 3316\n",
+ "3316 3322 0 0 + 6 3322\n",
+ "3322 3323 0 0 + 1 3323\n",
+ "3323 3348 0 0 + 25 3348\n",
+ "3348 3349 0 0 + 1 3349\n",
+ "3349 3350 0 0 + 1 3350\n",
+ "3350 3351 0 0 + 1 3351\n",
+ "3351 3352 0 0 + 1 3352\n",
+ "3352 3353 0 0 + 1 3353\n",
+ "3353 3354 0 0 + 1 3354\n",
+ "3354 3356 0 0 + 2 3356\n",
+ "3356 3357 0 0 + 1 3357\n",
+ "3357 3489 0 0 + 132 3489\n",
+ "3489 3490 0 0 + 1 3490\n",
+ "3490 3642 0 0 + 152 3642\n",
+ "3642 3644 0 0 + 2 3644\n",
+ "3644 3685 0 0 + 41 3685\n",
+ "3685 3687 0 0 + 2 3687\n",
+ "3687 3693 0 0 + 6 3693\n",
+ "3693 3694 0 0 + 1 3694\n",
+ "3694 3708 0 0 + 14 3708\n",
+ "3708 3709 0 0 + 1 3709\n",
+ "3709 3710 0 0 + 1 3710\n",
+ "3710 3714 0 0 + 4 3714\n",
+ "3714 3715 0 0 + 1 3715\n",
+ "3715 3716 0 0 + 1 3716\n",
+ "3716 3720 0 0 + 4 3720\n",
+ "3720 3721 0 0 + 1 3721\n",
+ "3721 3722 0 0 + 1 3722\n",
+ "3722 3735 0 10 + 23 3735\n",
+ "[gaf2aln::Alignment position processing] Running on ALN_2 ...\n",
+ "0 1 0 0 + 1 1\n",
+ "1 2 0 0 + 1 2\n",
+ "2 3 0 0 + 1 3\n",
+ "3 4 0 0 + 1 4\n",
+ "4 5 0 0 + 1 5\n",
+ "5 6 0 0 + 1 6\n",
+ "6 7 0 0 + 1 7\n",
+ "7 8 0 0 + 1 8\n",
+ "8 9 0 0 + 1 9\n",
+ "9 10 0 0 + 1 10\n",
+ "10 11 0 0 + 1 11\n",
+ "11 12 0 0 + 1 12\n",
+ "12 13 0 0 + 1 13\n",
+ "13 14 0 0 + 1 14\n",
+ "14 15 0 0 + 1 15\n",
+ "15 16 0 0 + 1 16\n",
+ "16 17 0 0 + 1 17\n",
+ "17 18 0 0 + 1 18\n",
+ "18 19 0 0 + 1 19\n",
+ "19 20 0 0 + 1 20\n",
+ "20 21 0 0 + 1 21\n",
+ "21 22 0 0 + 1 22\n",
+ "22 23 0 0 + 1 23\n",
+ "23 24 0 0 + 1 24\n",
+ "24 25 0 0 + 1 25\n",
+ "25 26 0 0 + 1 26\n",
+ "26 27 0 0 + 1 27\n",
+ "27 28 0 0 + 1 28\n",
+ "28 29 0 0 + 1 29\n",
+ "29 30 0 0 + 1 30\n",
+ "30 31 0 0 + 1 31\n",
+ "31 32 0 0 + 1 32\n",
+ "32 33 0 0 + 1 33\n",
+ "33 34 0 0 + 1 34\n",
+ "34 35 0 0 + 1 35\n",
+ "35 36 0 0 + 1 36\n",
+ "36 37 0 0 + 1 37\n",
+ "37 38 0 0 + 1 38\n",
+ "38 39 0 0 + 1 39\n",
+ "39 40 0 0 + 1 40\n",
+ "40 41 0 0 + 1 41\n",
+ "41 42 0 0 + 1 42\n",
+ "42 43 0 0 + 1 43\n",
+ "43 44 0 0 + 1 44\n",
+ "44 45 0 0 + 1 45\n",
+ "45 46 0 0 + 1 46\n",
+ "46 47 0 0 + 1 47\n",
+ "47 48 0 0 + 1 48\n",
+ "48 49 0 0 + 1 49\n",
+ "49 50 0 0 + 1 50\n",
+ "50 51 0 0 + 1 51\n",
+ "51 52 0 0 + 1 52\n",
+ "52 53 0 0 + 1 53\n",
+ "53 54 0 0 + 1 54\n",
+ "54 55 0 0 + 1 55\n",
+ "55 56 0 0 + 1 56\n",
+ "56 57 0 0 + 1 57\n",
+ "57 58 0 0 + 1 58\n",
+ "58 59 0 0 + 1 59\n",
+ "59 60 0 0 + 1 60\n",
+ "60 61 0 0 + 1 61\n",
+ "61 62 0 0 + 1 62\n",
+ "62 63 0 0 + 1 63\n",
+ "63 64 0 0 + 1 64\n",
+ "64 65 0 0 + 1 65\n",
+ "65 66 0 0 + 1 66\n",
+ "66 67 0 0 + 1 67\n",
+ "67 68 0 0 + 1 68\n",
+ "68 69 0 0 + 1 69\n",
+ "69 70 0 0 + 1 70\n",
+ "70 71 0 0 + 1 71\n",
+ "71 72 0 0 + 1 72\n",
+ "72 73 0 0 + 1 73\n",
+ "73 74 0 0 + 1 74\n",
+ "74 75 0 0 + 1 75\n",
+ "75 76 0 0 + 1 76\n",
+ "76 77 0 0 + 1 77\n",
+ "77 78 0 0 + 1 78\n",
+ "78 3735 0 57489 + 61146 3735\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(f\"[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...\")\n",
+ "# Adding nodes positions relative to path\n",
+ "\n",
+ "def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length):\n",
+ " # Initializing current position in query\n",
+ " cur_pos = 0\n",
+ "\n",
+ " # Getting start and end node ids\n",
+ " start_end_id = (aln_dict[aln_name][\"PATH.MATCH\"][0][0], aln_dict[aln_name][\"PATH.MATCH\"][-1][0])\n",
+ "\n",
+ " # Creating result dictionnary\n",
+ " res = {}\n",
+ "\n",
+ " ## Iterating over node_ids from the given alignment\n",
+ " for node_id, orient in aln_dict[aln_name][\"PATH.MATCH\"]:\n",
+ " # Adding entry for current node\n",
+ " res[node_id] = {aln_name: {}}\n",
+ "\n",
+ " # First node\n",
+ " if node_id == start_end_id[0]:\n",
+ " start_pos = 0\n",
+ " s_off = int(aln_dict[aln_name][\"ALN.START\"])\n",
+ " end_pos = nodes_length[node_id]-s_off\n",
+ " e_off = 0\n",
+ " # End node\n",
+ " elif node_id == start_end_id[1]:\n",
+ " start_pos = cur_pos\n",
+ " s_off = 0\n",
+ " end_pos = int(aln_dict[aln_name][\"QRY.END\"])\n",
+ " e_off = nodes_length[node_id]-(end_pos-cur_pos)\n",
+ " # Node in between\n",
+ " else :\n",
+ " start_pos = cur_pos\n",
+ " s_off, e_off = 0, 0\n",
+ " end_pos = cur_pos+nodes_length[node_id]\n",
+ "\n",
+ " res[node_id] = {\n",
+ " \"START\": start_pos, # Start position on the query\n",
+ " \"END\": end_pos, # End position on the query\n",
+ " \"S.OFF\": s_off, # Offset between the start of the alignment and the node's start\n",
+ " \"E.OFF\": e_off, # Offset between the end of the alignment and the node's end \n",
+ " \"STRAND\": orient # Orientation of the node in the alignment\n",
+ " }\n",
+ " \n",
+ " cur_pos = end_pos\n",
+ " print(start_pos, end_pos, s_off, e_off, orient, nodes_length[node_id], cur_pos)\n",
+ "\n",
+ " return res\n",
+ "\n",
+ "# Storing alignement \n",
+ "res = {}\n",
+ "executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)\n",
+ "for aln_name in aln_dict.keys():\n",
+ " print(f\"[gaf2aln::Alignment position processing] Running on {aln_name} ...\")\n",
+ " \n",
+ " res[aln_name] = executor.submit(get_aln_node_info, aln_name)\n",
+ " #res[aln_name] = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)\n",
+ "\n",
+ "executor.shutdown(wait=True)\n",
+ "\n",
+ "for aln_name, node_info in res.items():\n",
+ " results = node_info.result()\n",
+ " for node_id, info in results.items():\n",
+ " nodes[node_id][aln_name] = info\n",
+ "\n",
+ "del res"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 7,
+ "id": "4c30727c-7ffc-4852-ad81-ca2a5a7f9957",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...\n",
+ "[gaf2aln::CIGAR processing] Running on ALN_1 ...\n",
+ "[gaf2aln::CIGAR processing] Running on ALN_2 ...\n"
+ ]
+ }
+ ],
+ "source": [
+ "# Calculating CIGAR for each nodes in each aln\n",
+ "print(f\"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...\")\n",
+ "# Iterating over alignments\n",
+ "for aln in aln_dict.keys():\n",
+ " \n",
+ " print(f\"[gaf2aln::CIGAR processing] Running on {aln} ...\")\n",
+ " # Getting the list of base level alignement ([\"=\", \"X\", ...] from \"1=1X...\")\n",
+ " raw_cigar = cigar2basealn(aln_dict[aln][\"RAW.CIGAR\"])\n",
+ " CIGAR={}\n",
+ "\n",
+ " for node_id, orient in aln_dict[aln][\"PATH.MATCH\"]:\n",
+ "\n",
+ " _cigar = basealn2cigar(raw_cigar[\n",
+ " nodes[node_id][aln][\"START\"]:nodes[node_id][aln][\"END\"]\n",
+ " ])\n",
+ " nodes[node_id][aln][\"CIGAR\"] = _cigar\n",
+ " #print(_cigar, nodes[node_id][aln][\"START\"], nodes[node_id][aln][\"END\"])"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 8,
+ "id": "e15e4762-cd71-4afe-bc74-ebe44869fee6",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "ALN_1\n",
+ "7046526 D134#1#chr03 0 77\n",
+ "\tIn path\n",
+ "\t 73306158 73306235\n",
+ "skipped\n",
+ "\n",
+ "7046528 D134#1#chr03 77 82\n",
+ "\tIn path\n",
+ "\t 73306238 73306243\n",
+ "{'Q.START': 0, 'Q.END': 77, 'T.START': 73306158, 'T.END': 73306235, 'CG': '77='}\n",
+ "7046530 D134#1#chr03 82 83\n",
+ "\tNot in path\n",
+ "7046531 D134#1#chr03 83 138\n",
+ "\tIn path\n",
+ "\t 73306246 73306301\n",
+ "{'Q.START': 77, 'Q.END': 82, 'T.START': 73306238, 'T.END': 73306243, 'CG': '5='}\n",
+ "7046532 D134#1#chr03 138 139\n",
+ "\tNot in path\n",
+ "7046533 D134#1#chr03 139 202\n",
+ "\tIn path\n",
+ "\t 73306302 73306365\n",
+ "{'Q.START': 83, 'Q.END': 138, 'T.START': 73306246, 'T.END': 73306301, 'CG': '55='}\n",
+ "7046534 D134#1#chr03 202 203\n",
+ "\tIn path\n",
+ "\t 73306366 73306367\n",
+ "{'Q.START': 139, 'Q.END': 202, 'T.START': 73306302, 'T.END': 73306365, 'CG': '63='}\n",
+ "7046536 D134#1#chr03 203 379\n",
+ "\tIn path\n",
+ "\t 73306368 73306544\n",
+ "{'Q.START': 202, 'Q.END': 203, 'T.START': 73306366, 'T.END': 73306367, 'CG': '1='}\n",
+ "7046537 D134#1#chr03 379 380\n",
+ "\tIn path\n",
+ "\t 73306545 73306546\n",
+ "{'Q.START': 203, 'Q.END': 379, 'T.START': 73306368, 'T.END': 73306544, 'CG': '176='}\n",
+ "7046539 D134#1#chr03 380 429\n",
+ "\tIn path\n",
+ "\t 73306547 73306596\n",
+ "{'Q.START': 379, 'Q.END': 380, 'T.START': 73306545, 'T.END': 73306546, 'CG': '1='}\n",
+ "7046541 D134#1#chr03 429 430\n",
+ "\tIn path\n",
+ "\t 73306597 73306598\n",
+ "{'Q.START': 380, 'Q.END': 429, 'T.START': 73306547, 'T.END': 73306596, 'CG': '49='}\n",
+ "7046542 D134#1#chr03 430 457\n",
+ "\tIn path\n",
+ "\t 73306599 73306626\n",
+ "{'Q.START': 429, 'Q.END': 430, 'T.START': 73306597, 'T.END': 73306598, 'CG': '1='}\n",
+ "7046544 D134#1#chr03 457 492\n",
+ "\tIn path\n",
+ "\t 73306641 73306676\n",
+ "{'Q.START': 430, 'Q.END': 457, 'T.START': 73306599, 'T.END': 73306626, 'CG': '27='}\n",
+ "7046546 D134#1#chr03 492 494\n",
+ "\tNot in path\n",
+ "7046547 D134#1#chr03 494 497\n",
+ "\tNot in path\n",
+ "7046549 D134#1#chr03 497 507\n",
+ "\tNot in path\n",
+ "7046551 D134#1#chr03 507 508\n",
+ "\tNot in path\n",
+ "7046552 D134#1#chr03 508 564\n",
+ "\tIn path\n",
+ "\t 73306694 73306750\n",
+ "{'Q.START': 457, 'Q.END': 492, 'T.START': 73306641, 'T.END': 73306676, 'CG': '35='}\n",
+ "7046554 D134#1#chr03 564 566\n",
+ "\tNot in path\n",
+ "7046556 D134#1#chr03 568 569\n",
+ "\tIn path\n",
+ "\t 73306753 73306754\n",
+ "{'Q.START': 508, 'Q.END': 564, 'T.START': 73306694, 'T.END': 73306750, 'CG': '56='}\n",
+ "7046556 D134#1#chr03 568 569\n",
+ "\tIn path\n",
+ "\t 73306753 73306754\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
+ "7046556 D134#1#chr03 568 569\n",
+ "\tIn path\n",
+ "\t 73306753 73306754\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
+ "7046557 D134#1#chr03 569 824\n",
+ "\tIn path\n",
+ "\t 73306755 73307010\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
+ "7046558 D134#1#chr03 824 826\n",
+ "\tNot in path\n",
+ "7046559 D134#1#chr03 826 858\n",
+ "\tIn path\n",
+ "\t 73307011 73307043\n",
+ "{'Q.START': 569, 'Q.END': 824, 'T.START': 73306755, 'T.END': 73307010, 'CG': '255='}\n",
+ "7046560 D134#1#chr03 858 859\n",
+ "\tIn path\n",
+ "\t 73307044 73307045\n",
+ "{'Q.START': 826, 'Q.END': 858, 'T.START': 73307011, 'T.END': 73307043, 'CG': '32='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 858, 'Q.END': 859, 'T.START': 73307044, 'T.END': 73307045, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046561 D134#1#chr03 868 869\n",
+ "\tIn path\n",
+ "\t 73307046 73307047\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046562 D134#1#chr03 869 913\n",
+ "\tIn path\n",
+ "\t 73307048 73307092\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
+ "7046564 D134#1#chr03 913 919\n",
+ "\tIn path\n",
+ "\t 73307093 73307099\n",
+ "{'Q.START': 869, 'Q.END': 913, 'T.START': 73307048, 'T.END': 73307092, 'CG': '44='}\n",
+ "7046565 D134#1#chr03 919 978\n",
+ "\tIn path\n",
+ "\t 73307100 73307159\n",
+ "{'Q.START': 913, 'Q.END': 919, 'T.START': 73307093, 'T.END': 73307099, 'CG': '6='}\n",
+ "7046567 D134#1#chr03 978 979\n",
+ "\tIn path\n",
+ "\t 73307160 73307161\n",
+ "{'Q.START': 919, 'Q.END': 978, 'T.START': 73307100, 'T.END': 73307159, 'CG': '59='}\n",
+ "7046568 D134#1#chr03 979 1038\n",
+ "\tIn path\n",
+ "\t 73307162 73307221\n",
+ "{'Q.START': 978, 'Q.END': 979, 'T.START': 73307160, 'T.END': 73307161, 'CG': '1='}\n",
+ "7046570 D134#1#chr03 1038 1045\n",
+ "\tIn path\n",
+ "\t 73307224 73307231\n",
+ "{'Q.START': 979, 'Q.END': 1038, 'T.START': 73307162, 'T.END': 73307221, 'CG': '59='}\n",
+ "7046571 D134#1#chr03 1045 1046\n",
+ "\tIn path\n",
+ "\t 73307232 73307233\n",
+ "{'Q.START': 1038, 'Q.END': 1045, 'T.START': 73307224, 'T.END': 73307231, 'CG': '7='}\n",
+ "7046573 D134#1#chr03 1046 1080\n",
+ "\tIn path\n",
+ "\t 73307234 73307268\n",
+ "{'Q.START': 1045, 'Q.END': 1046, 'T.START': 73307232, 'T.END': 73307233, 'CG': '1='}\n",
+ "7046574 D134#1#chr03 1080 1081\n",
+ "\tIn path\n",
+ "\t 73307269 73307270\n",
+ "{'Q.START': 1046, 'Q.END': 1080, 'T.START': 73307234, 'T.END': 73307268, 'CG': '34='}\n",
+ "7046576 D134#1#chr03 1081 1107\n",
+ "\tIn path\n",
+ "\t 73307271 73307297\n",
+ "{'Q.START': 1080, 'Q.END': 1081, 'T.START': 73307269, 'T.END': 73307270, 'CG': '1='}\n",
+ "7046577 D134#1#chr03 1107 1108\n",
+ "\tNot in path\n",
+ "7046579 D134#1#chr03 1108 1183\n",
+ "\tIn path\n",
+ "\t 73307300 73307375\n",
+ "{'Q.START': 1081, 'Q.END': 1107, 'T.START': 73307271, 'T.END': 73307297, 'CG': '26='}\n",
+ "7046581 D134#1#chr03 1183 1186\n",
+ "\tIn path\n",
+ "\t 73307376 73307379\n",
+ "{'Q.START': 1108, 'Q.END': 1183, 'T.START': 73307300, 'T.END': 73307375, 'CG': '75='}\n",
+ "7046583 D134#1#chr03 1186 1224\n",
+ "\tNot in path\n",
+ "7046584 D134#1#chr03 1224 1257\n",
+ "\tIn path\n",
+ "\t 73307419 73307452\n",
+ "{'Q.START': 1183, 'Q.END': 1186, 'T.START': 73307376, 'T.END': 73307379, 'CG': '3='}\n",
+ "7046586 D134#1#chr03 1257 1289\n",
+ "\tNot in path\n",
+ "7046587 D134#1#chr03 1289 1311\n",
+ "\tIn path\n",
+ "\t 73307475 73307497\n",
+ "{'Q.START': 1224, 'Q.END': 1257, 'T.START': 73307419, 'T.END': 73307452, 'CG': '33='}\n",
+ "7046589 D134#1#chr03 1311 1359\n",
+ "\tNot in path\n",
+ "7046590 D134#1#chr03 1359 1382\n",
+ "\tIn path\n",
+ "\t 73307546 73307569\n",
+ "{'Q.START': 1289, 'Q.END': 1311, 'T.START': 73307475, 'T.END': 73307497, 'CG': '22='}\n",
+ "7046592 D134#1#chr03 1382 1434\n",
+ "\tNot in path\n",
+ "7046593 D134#1#chr03 1434 1451\n",
+ "\tIn path\n",
+ "\t 73307643 73307660\n",
+ "{'Q.START': 1359, 'Q.END': 1382, 'T.START': 73307546, 'T.END': 73307569, 'CG': '23='}\n",
+ "7046594 D134#1#chr03 1451 1531\n",
+ "\tIn path\n",
+ "\t 73307661 73307741\n",
+ "{'Q.START': 1434, 'Q.END': 1451, 'T.START': 73307643, 'T.END': 73307660, 'CG': '17='}\n",
+ "7046596 D134#1#chr03 1531 1532\n",
+ "\tNot in path\n",
+ "7046597 D134#1#chr03 1532 1543\n",
+ "\tIn path\n",
+ "\t 73307744 73307755\n",
+ "{'Q.START': 1451, 'Q.END': 1531, 'T.START': 73307661, 'T.END': 73307741, 'CG': '80='}\n",
+ "7046599 D134#1#chr03 1543 1544\n",
+ "\tNot in path\n",
+ "7046600 D134#1#chr03 1544 1572\n",
+ "\tIn path\n",
+ "\t 73307758 73307786\n",
+ "{'Q.START': 1532, 'Q.END': 1543, 'T.START': 73307744, 'T.END': 73307755, 'CG': '11='}\n",
+ "7046601 D134#1#chr03 1572 1573\n",
+ "\tIn path\n",
+ "\t 73307787 73307788\n",
+ "{'Q.START': 1544, 'Q.END': 1572, 'T.START': 73307758, 'T.END': 73307786, 'CG': '28='}\n",
+ "7046603 D134#1#chr03 1573 1587\n",
+ "\tIn path\n",
+ "\t 73307789 73307803\n",
+ "{'Q.START': 1572, 'Q.END': 1573, 'T.START': 73307787, 'T.END': 73307788, 'CG': '1='}\n",
+ "7046604 D134#1#chr03 1587 1588\n",
+ "\tNot in path\n",
+ "7046606 D134#1#chr03 1588 1616\n",
+ "\tIn path\n",
+ "\t 73307806 73307834\n",
+ "{'Q.START': 1573, 'Q.END': 1587, 'T.START': 73307789, 'T.END': 73307803, 'CG': '14='}\n",
+ "7046608 D134#1#chr03 1616 1617\n",
+ "\tIn path\n",
+ "\t 73307835 73307836\n",
+ "{'Q.START': 1588, 'Q.END': 1616, 'T.START': 73307806, 'T.END': 73307834, 'CG': '28='}\n",
+ "7046609 D134#1#chr03 1617 1646\n",
+ "\tIn path\n",
+ "\t 73307837 73307866\n",
+ "{'Q.START': 1616, 'Q.END': 1617, 'T.START': 73307835, 'T.END': 73307836, 'CG': '1='}\n",
+ "7046621 D134#1#chr03 1646 1661\n",
+ "\tIn path\n",
+ "\t 73307867 73307882\n",
+ "{'Q.START': 1617, 'Q.END': 1646, 'T.START': 73307837, 'T.END': 73307866, 'CG': '29='}\n",
+ "7046622 D134#1#chr03 1661 1673\n",
+ "\tIn path\n",
+ "\t 73307883 73307895\n",
+ "{'Q.START': 1646, 'Q.END': 1661, 'T.START': 73307867, 'T.END': 73307882, 'CG': '15='}\n",
+ "7046624 D134#1#chr03 1673 1674\n",
+ "\tIn path\n",
+ "\t 73307896 73307897\n",
+ "{'Q.START': 1661, 'Q.END': 1673, 'T.START': 73307883, 'T.END': 73307895, 'CG': '12='}\n",
+ "7046625 D134#1#chr03 1674 1726\n",
+ "\tIn path\n",
+ "\t 73307898 73307950\n",
+ "{'Q.START': 1673, 'Q.END': 1674, 'T.START': 73307896, 'T.END': 73307897, 'CG': '1='}\n",
+ "7046626 D134#1#chr03 1726 1727\n",
+ "\tNot in path\n",
+ "7046628 D134#1#chr03 1727 1762\n",
+ "\tIn path\n",
+ "\t 73307953 73307988\n",
+ "{'Q.START': 1674, 'Q.END': 1726, 'T.START': 73307898, 'T.END': 73307950, 'CG': '52='}\n",
+ "7046631 D134#1#chr03 1766 1767\n",
+ "\tIn path\n",
+ "\t 73307991 73307992\n",
+ "{'Q.START': 1727, 'Q.END': 1762, 'T.START': 73307953, 'T.END': 73307988, 'CG': '35='}\n",
+ "7046673 D134#1#chr03 1765 1766\n",
+ "\tIn path\n",
+ "\t 73307993 73307994\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
+ "7046631 D134#1#chr03 1766 1767\n",
+ "\tIn path\n",
+ "\t 73307991 73307992\n",
+ "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}\n",
+ "7046673 D134#1#chr03 1765 1766\n",
+ "\tIn path\n",
+ "\t 73307993 73307994\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
+ "7046631 D134#1#chr03 1766 1767\n",
+ "\tIn path\n",
+ "\t 73307991 73307992\n",
+ "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}\n",
+ "7046632 D134#1#chr03 1767 1824\n",
+ "\tIn path\n",
+ "\t 73307995 73308052\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
+ "7046634 D134#1#chr03 1824 1825\n",
+ "\tIn path\n",
+ "\t 73308053 73308054\n",
+ "{'Q.START': 1767, 'Q.END': 1824, 'T.START': 73307995, 'T.END': 73308052, 'CG': '57='}\n",
+ "7046635 D134#1#chr03 1825 1975\n",
+ "\tIn path\n",
+ "\t 73308055 73308205\n",
+ "{'Q.START': 1824, 'Q.END': 1825, 'T.START': 73308053, 'T.END': 73308054, 'CG': '1='}\n",
+ "7046637 D134#1#chr03 1975 1976\n",
+ "\tNot in path\n",
+ "7046638 D134#1#chr03 1976 2015\n",
+ "\tIn path\n",
+ "\t 73308208 73308247\n",
+ "{'Q.START': 1825, 'Q.END': 1975, 'T.START': 73308055, 'T.END': 73308205, 'CG': '150='}\n",
+ "7046639 D134#1#chr03 2015 2016\n",
+ "\tNot in path\n",
+ "7046641 D134#1#chr03 2016 2047\n",
+ "\tIn path\n",
+ "\t 73308250 73308281\n",
+ "{'Q.START': 1976, 'Q.END': 2015, 'T.START': 73308208, 'T.END': 73308247, 'CG': '39='}\n",
+ "7046644 D134#1#chr03 2047 2055\n",
+ "\tIn path\n",
+ "\t 73308286 73308294\n",
+ "{'Q.START': 2016, 'Q.END': 2047, 'T.START': 73308250, 'T.END': 73308281, 'CG': '31='}\n",
+ "7046646 D134#1#chr03 2055 2056\n",
+ "\tNot in path\n",
+ "7046647 D134#1#chr03 2056 2120\n",
+ "\tIn path\n",
+ "\t 73308297 73308361\n",
+ "{'Q.START': 2047, 'Q.END': 2055, 'T.START': 73308286, 'T.END': 73308294, 'CG': '8='}\n",
+ "7046649 D134#1#chr03 2120 2121\n",
+ "\tIn path\n",
+ "\t 73308362 73308363\n",
+ "{'Q.START': 2056, 'Q.END': 2120, 'T.START': 73308297, 'T.END': 73308361, 'CG': '64='}\n",
+ "7046650 D134#1#chr03 2121 2157\n",
+ "\tIn path\n",
+ "\t 73308364 73308400\n",
+ "{'Q.START': 2120, 'Q.END': 2121, 'T.START': 73308362, 'T.END': 73308363, 'CG': '1='}\n",
+ "7046652 D134#1#chr03 2157 2158\n",
+ "\tNot in path\n",
+ "7046653 D134#1#chr03 2158 2170\n",
+ "\tIn path\n",
+ "\t 73308403 73308415\n",
+ "{'Q.START': 2121, 'Q.END': 2157, 'T.START': 73308364, 'T.END': 73308400, 'CG': '36='}\n",
+ "7046654 D134#1#chr03 2170 2171\n",
+ "\tIn path\n",
+ "\t 73308416 73308417\n",
+ "{'Q.START': 2158, 'Q.END': 2170, 'T.START': 73308403, 'T.END': 73308415, 'CG': '12='}\n",
+ "7046656 D134#1#chr03 2171 2205\n",
+ "\tIn path\n",
+ "\t 73308418 73308452\n",
+ "{'Q.START': 2170, 'Q.END': 2171, 'T.START': 73308416, 'T.END': 73308417, 'CG': '1='}\n",
+ "7046657 D134#1#chr03 2205 2206\n",
+ "\tNot in path\n",
+ "7046659 D134#1#chr03 2206 2344\n",
+ "\tIn path\n",
+ "\t 73308455 73308593\n",
+ "{'Q.START': 2171, 'Q.END': 2205, 'T.START': 73308418, 'T.END': 73308452, 'CG': '34='}\n",
+ "7046660 D134#1#chr03 2344 2345\n",
+ "\tNot in path\n",
+ "7046662 D134#1#chr03 2345 2364\n",
+ "\tIn path\n",
+ "\t 73308596 73308615\n",
+ "{'Q.START': 2206, 'Q.END': 2344, 'T.START': 73308455, 'T.END': 73308593, 'CG': '138='}\n",
+ "7046663 D134#1#chr03 2364 2383\n",
+ "\tIn path\n",
+ "\t 73308616 73308635\n",
+ "{'Q.START': 2345, 'Q.END': 2364, 'T.START': 73308596, 'T.END': 73308615, 'CG': '19='}\n",
+ "7046665 D134#1#chr03 2383 2408\n",
+ "\tIn path\n",
+ "\t 73308636 73308661\n",
+ "{'Q.START': 2364, 'Q.END': 2383, 'T.START': 73308616, 'T.END': 73308635, 'CG': '19='}\n",
+ "7046667 D134#1#chr03 2408 2409\n",
+ "\tIn path\n",
+ "\t 73308662 73308663\n",
+ "{'Q.START': 2383, 'Q.END': 2408, 'T.START': 73308636, 'T.END': 73308661, 'CG': '25='}\n",
+ "7046668 D134#1#chr03 2409 2441\n",
+ "\tIn path\n",
+ "\t 73308664 73308696\n",
+ "{'Q.START': 2408, 'Q.END': 2409, 'T.START': 73308662, 'T.END': 73308663, 'CG': '1='}\n",
+ "7046670 D134#1#chr03 2441 2442\n",
+ "\tIn path\n",
+ "\t 73308697 73308698\n",
+ "{'Q.START': 2409, 'Q.END': 2441, 'T.START': 73308664, 'T.END': 73308696, 'CG': '32='}\n",
+ "7046671 D134#1#chr03 2442 2580\n",
+ "\tIn path\n",
+ "\t 73308699 73308837\n",
+ "{'Q.START': 2441, 'Q.END': 2442, 'T.START': 73308697, 'T.END': 73308698, 'CG': '1='}\n",
+ "7046674 D134#1#chr03 2582 2583\n",
+ "\tIn path\n",
+ "\t 73308838 73308839\n",
+ "{'Q.START': 2442, 'Q.END': 2580, 'T.START': 73308699, 'T.END': 73308837, 'CG': '138='}\n",
+ "7046675 D134#1#chr03 2583 2584\n",
+ "\tIn path\n",
+ "\t 73308840 73308841\n",
+ "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}\n",
+ "7046674 D134#1#chr03 2582 2583\n",
+ "\tIn path\n",
+ "\t 73308838 73308839\n",
+ "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}\n",
+ "7046675 D134#1#chr03 2583 2584\n",
+ "\tIn path\n",
+ "\t 73308840 73308841\n",
+ "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}\n",
+ "7046676 D134#1#chr03 2584 2764\n",
+ "\tIn path\n",
+ "\t 73308842 73309022\n",
+ "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}\n",
+ "7046678 D134#1#chr03 2764 2765\n",
+ "\tNot in path\n",
+ "7046679 D134#1#chr03 2765 2797\n",
+ "\tIn path\n",
+ "\t 73309025 73309057\n",
+ "{'Q.START': 2584, 'Q.END': 2764, 'T.START': 73308842, 'T.END': 73309022, 'CG': '180='}\n",
+ "7046680 D134#1#chr03 2797 2798\n",
+ "\tNot in path\n",
+ "7046682 D134#1#chr03 2798 2878\n",
+ "\tIn path\n",
+ "\t 73309060 73309140\n",
+ "{'Q.START': 2765, 'Q.END': 2797, 'T.START': 73309025, 'T.END': 73309057, 'CG': '32='}\n",
+ "7046684 D134#1#chr03 2878 2879\n",
+ "\tIn path\n",
+ "\t 73309141 73309142\n",
+ "{'Q.START': 2798, 'Q.END': 2878, 'T.START': 73309060, 'T.END': 73309140, 'CG': '80='}\n",
+ "7046685 D134#1#chr03 2879 2951\n",
+ "\tIn path\n",
+ "\t 73309143 73309215\n",
+ "{'Q.START': 2878, 'Q.END': 2879, 'T.START': 73309141, 'T.END': 73309142, 'CG': '1='}\n",
+ "7046686 D134#1#chr03 2951 2952\n",
+ "\tIn path\n",
+ "\t 73309216 73309217\n",
+ "{'Q.START': 2879, 'Q.END': 2951, 'T.START': 73309143, 'T.END': 73309215, 'CG': '72='}\n",
+ "7046688 D134#1#chr03 2952 3002\n",
+ "\tIn path\n",
+ "\t 73309218 73309268\n",
+ "{'Q.START': 2951, 'Q.END': 2952, 'T.START': 73309216, 'T.END': 73309217, 'CG': '1='}\n",
+ "7046690 D134#1#chr03 3002 3077\n",
+ "\tIn path\n",
+ "\t 73309271 73309346\n",
+ "{'Q.START': 2952, 'Q.END': 3002, 'T.START': 73309218, 'T.END': 73309268, 'CG': '50='}\n",
+ "7046692 D134#1#chr03 3077 3078\n",
+ "\tIn path\n",
+ "\t 73309347 73309348\n",
+ "{'Q.START': 3002, 'Q.END': 3077, 'T.START': 73309271, 'T.END': 73309346, 'CG': '75='}\n",
+ "7046693 D134#1#chr03 3078 3093\n",
+ "\tIn path\n",
+ "\t 73309349 73309364\n",
+ "{'Q.START': 3077, 'Q.END': 3078, 'T.START': 73309347, 'T.END': 73309348, 'CG': '1='}\n",
+ "7046695 D134#1#chr03 3093 3094\n",
+ "\tNot in path\n",
+ "7046696 D134#1#chr03 3094 3097\n",
+ "\tIn path\n",
+ "\t 73309367 73309370\n",
+ "{'Q.START': 3078, 'Q.END': 3093, 'T.START': 73309349, 'T.END': 73309364, 'CG': '15='}\n",
+ "7046698 D134#1#chr03 3097 3140\n",
+ "\tIn path\n",
+ "\t 73309371 73309414\n",
+ "{'Q.START': 3094, 'Q.END': 3097, 'T.START': 73309367, 'T.END': 73309370, 'CG': '3='}\n",
+ "7046700 D134#1#chr03 3140 3210\n",
+ "\tIn path\n",
+ "\t 73309415 73309485\n",
+ "{'Q.START': 3097, 'Q.END': 3140, 'T.START': 73309371, 'T.END': 73309414, 'CG': '43='}\n",
+ "7046702 D134#1#chr03 3210 3211\n",
+ "\tIn path\n",
+ "\t 73309486 73309487\n",
+ "{'Q.START': 3140, 'Q.END': 3210, 'T.START': 73309415, 'T.END': 73309485, 'CG': '70='}\n",
+ "7046703 D134#1#chr03 3211 3229\n",
+ "\tIn path\n",
+ "\t 73309488 73309506\n",
+ "{'Q.START': 3210, 'Q.END': 3211, 'T.START': 73309486, 'T.END': 73309487, 'CG': '1='}\n",
+ "7046704 D134#1#chr03 3229 3230\n",
+ "\tIn path\n",
+ "\t 73309507 73309508\n",
+ "{'Q.START': 3211, 'Q.END': 3229, 'T.START': 73309488, 'T.END': 73309506, 'CG': '18='}\n",
+ "7046706 D134#1#chr03 3230 3276\n",
+ "\tIn path\n",
+ "\t 73309509 73309555\n",
+ "{'Q.START': 3229, 'Q.END': 3230, 'T.START': 73309507, 'T.END': 73309508, 'CG': '1='}\n",
+ "7046707 D134#1#chr03 3276 3277\n",
+ "\tNot in path\n",
+ "7046709 D134#1#chr03 3277 3315\n",
+ "\tIn path\n",
+ "\t 73309558 73309596\n",
+ "{'Q.START': 3230, 'Q.END': 3276, 'T.START': 73309509, 'T.END': 73309555, 'CG': '46='}\n",
+ "7046710 D134#1#chr03 3315 3316\n",
+ "\tNot in path\n",
+ "7046712 D134#1#chr03 3316 3322\n",
+ "\tIn path\n",
+ "\t 73309599 73309605\n",
+ "{'Q.START': 3277, 'Q.END': 3315, 'T.START': 73309558, 'T.END': 73309596, 'CG': '38='}\n",
+ "7046713 D134#1#chr03 3322 3323\n",
+ "\tNot in path\n",
+ "7046715 D134#1#chr03 3323 3348\n",
+ "\tIn path\n",
+ "\t 73309608 73309633\n",
+ "{'Q.START': 3316, 'Q.END': 3322, 'T.START': 73309599, 'T.END': 73309605, 'CG': '6='}\n",
+ "7046718 D134#1#chr03 3352 3353\n",
+ "\tIn path\n",
+ "\t 73309634 73309635\n",
+ "{'Q.START': 3323, 'Q.END': 3348, 'T.START': 73309608, 'T.END': 73309633, 'CG': '25='}\n",
+ "7046717 D134#1#chr03 3351 3352\n",
+ "\tIn path\n",
+ "\t 73309636 73309637\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
+ "7046718 D134#1#chr03 3352 3353\n",
+ "\tIn path\n",
+ "\t 73309634 73309635\n",
+ "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}\n",
+ "7046717 D134#1#chr03 3351 3352\n",
+ "\tIn path\n",
+ "\t 73309636 73309637\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
+ "7046718 D134#1#chr03 3352 3353\n",
+ "\tIn path\n",
+ "\t 73309634 73309635\n",
+ "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}\n",
+ "7046720 D134#1#chr03 3353 3354\n",
+ "\tIn path\n",
+ "\t 73309638 73309639\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
+ "7046722 D134#1#chr03 3354 3356\n",
+ "\tIn path\n",
+ "\t 73309640 73309642\n",
+ "{'Q.START': 3353, 'Q.END': 3354, 'T.START': 73309638, 'T.END': 73309639, 'CG': '1='}\n",
+ "7046724 D134#1#chr03 3356 3357\n",
+ "\tNot in path\n",
+ "7046725 D134#1#chr03 3357 3489\n",
+ "\tIn path\n",
+ "\t 73309645 73309777\n",
+ "{'Q.START': 3354, 'Q.END': 3356, 'T.START': 73309640, 'T.END': 73309642, 'CG': '2='}\n",
+ "7046727 D134#1#chr03 3489 3490\n",
+ "\tNot in path\n",
+ "7046728 D134#1#chr03 3490 3642\n",
+ "\tIn path\n",
+ "\t 73309780 73309932\n",
+ "{'Q.START': 3357, 'Q.END': 3489, 'T.START': 73309645, 'T.END': 73309777, 'CG': '132='}\n",
+ "7046729 D134#1#chr03 3642 3644\n",
+ "\tNot in path\n",
+ "7046730 D134#1#chr03 3644 3685\n",
+ "\tIn path\n",
+ "\t 73309933 73309974\n",
+ "{'Q.START': 3490, 'Q.END': 3642, 'T.START': 73309780, 'T.END': 73309932, 'CG': '152='}\n",
+ "7046731 D134#1#chr03 3685 3687\n",
+ "\tNot in path\n",
+ "7046733 D134#1#chr03 3687 3693\n",
+ "\tIn path\n",
+ "\t 73309977 73309983\n",
+ "{'Q.START': 3644, 'Q.END': 3685, 'T.START': 73309933, 'T.END': 73309974, 'CG': '41='}\n",
+ "7046735 D134#1#chr03 3693 3694\n",
+ "\tNot in path\n",
+ "7046736 D134#1#chr03 3694 3708\n",
+ "\tIn path\n",
+ "\t 73309986 73310000\n",
+ "{'Q.START': 3687, 'Q.END': 3693, 'T.START': 73309977, 'T.END': 73309983, 'CG': '6='}\n",
+ "7046738 D134#1#chr03 3720 3721\n",
+ "\tIn path\n",
+ "\t 73310010 73310011\n",
+ "{'Q.START': 3694, 'Q.END': 3708, 'T.START': 73309986, 'T.END': 73310000, 'CG': '14='}\n",
+ "7046739 D134#1#chr03 3721 3722\n",
+ "\tIn path\n",
+ "\t 73310003 73310004\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
+ "7046740 D134#1#chr03 3716 3720\n",
+ "\tIn path\n",
+ "\t 73310005 73310009\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
+ "7046738 D134#1#chr03 3720 3721\n",
+ "\tIn path\n",
+ "\t 73310010 73310011\n",
+ "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}\n",
+ "7046739 D134#1#chr03 3721 3722\n",
+ "\tIn path\n",
+ "\t 73310003 73310004\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
+ "7046740 D134#1#chr03 3716 3720\n",
+ "\tIn path\n",
+ "\t 73310005 73310009\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
+ "7046738 D134#1#chr03 3720 3721\n",
+ "\tIn path\n",
+ "\t 73310010 73310011\n",
+ "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}\n",
+ "7046739 D134#1#chr03 3721 3722\n",
+ "\tIn path\n",
+ "\t 73310003 73310004\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
+ "7046741 D134#1#chr03 3722 3735\n",
+ "\tIn path\n",
+ "\t 73310012 73310045\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
+ "ALN_1\n",
+ "7046526 TO1000#1#chr03 0 77\n",
+ "\t 64684013 64684090\n",
+ "skipped\n",
+ "\n",
+ "7046528 TO1000#1#chr03 77 82\n",
+ "\t 64684091 64684096\n",
+ "{'Q.START': 0, 'Q.END': 77, 'T.START': 64684013, 'T.END': 64684090, 'CG': '77='}\n",
+ "7046530 TO1000#1#chr03 82 83\n",
+ "\t 64684097 64684098\n",
+ "{'Q.START': 77, 'Q.END': 82, 'T.START': 64684091, 'T.END': 64684096, 'CG': '5='}\n",
+ "7046531 TO1000#1#chr03 83 138\n",
+ "\t 64684099 64684154\n",
+ "{'Q.START': 82, 'Q.END': 83, 'T.START': 64684097, 'T.END': 64684098, 'CG': '1='}\n",
+ "7046532 TO1000#1#chr03 138 139\n",
+ "\t 64684155 64684156\n",
+ "{'Q.START': 83, 'Q.END': 138, 'T.START': 64684099, 'T.END': 64684154, 'CG': '55='}\n",
+ "7046533 TO1000#1#chr03 139 202\n",
+ "\t 64684157 64684220\n",
+ "{'Q.START': 138, 'Q.END': 139, 'T.START': 64684155, 'T.END': 64684156, 'CG': '1='}\n",
+ "7046534 TO1000#1#chr03 202 203\n",
+ "\t 64684221 64684222\n",
+ "{'Q.START': 139, 'Q.END': 202, 'T.START': 64684157, 'T.END': 64684220, 'CG': '63='}\n",
+ "7046536 TO1000#1#chr03 203 379\n",
+ "\t 64684223 64684399\n",
+ "{'Q.START': 202, 'Q.END': 203, 'T.START': 64684221, 'T.END': 64684222, 'CG': '1='}\n",
+ "7046537 TO1000#1#chr03 379 380\n",
+ "\t 64684400 64684401\n",
+ "{'Q.START': 203, 'Q.END': 379, 'T.START': 64684223, 'T.END': 64684399, 'CG': '176='}\n",
+ "7046539 TO1000#1#chr03 380 429\n",
+ "\t 64684402 64684451\n",
+ "{'Q.START': 379, 'Q.END': 380, 'T.START': 64684400, 'T.END': 64684401, 'CG': '1='}\n",
+ "7046541 TO1000#1#chr03 429 430\n",
+ "\t 64684452 64684453\n",
+ "{'Q.START': 380, 'Q.END': 429, 'T.START': 64684402, 'T.END': 64684451, 'CG': '49='}\n",
+ "7046542 TO1000#1#chr03 430 457\n",
+ "\t 64684454 64684481\n",
+ "{'Q.START': 429, 'Q.END': 430, 'T.START': 64684452, 'T.END': 64684453, 'CG': '1='}\n",
+ "7046544 TO1000#1#chr03 457 492\n",
+ "\t 64684482 64684517\n",
+ "{'Q.START': 430, 'Q.END': 457, 'T.START': 64684454, 'T.END': 64684481, 'CG': '27='}\n",
+ "7046546 TO1000#1#chr03 492 494\n",
+ "\t 64684518 64684520\n",
+ "{'Q.START': 457, 'Q.END': 492, 'T.START': 64684482, 'T.END': 64684517, 'CG': '35='}\n",
+ "7046547 TO1000#1#chr03 494 497\n",
+ "\t 64684521 64684524\n",
+ "{'Q.START': 492, 'Q.END': 494, 'T.START': 64684518, 'T.END': 64684520, 'CG': '2='}\n",
+ "7046549 TO1000#1#chr03 497 507\n",
+ "\t 64684525 64684535\n",
+ "{'Q.START': 494, 'Q.END': 497, 'T.START': 64684521, 'T.END': 64684524, 'CG': '3='}\n",
+ "7046551 TO1000#1#chr03 507 508\n",
+ "\t 64684536 64684537\n",
+ "{'Q.START': 497, 'Q.END': 507, 'T.START': 64684525, 'T.END': 64684535, 'CG': '10='}\n",
+ "7046552 TO1000#1#chr03 508 564\n",
+ "\t 64684538 64684594\n",
+ "{'Q.START': 507, 'Q.END': 508, 'T.START': 64684536, 'T.END': 64684537, 'CG': '1='}\n",
+ "7046554 TO1000#1#chr03 564 566\n",
+ "\t 64684595 64684597\n",
+ "{'Q.START': 508, 'Q.END': 564, 'T.START': 64684538, 'T.END': 64684594, 'CG': '56='}\n",
+ "7046556 TO1000#1#chr03 568 569\n",
+ "\t 64684598 64684599\n",
+ "{'Q.START': 564, 'Q.END': 566, 'T.START': 64684595, 'T.END': 64684597, 'CG': '2='}\n",
+ "7046556 TO1000#1#chr03 568 569\n",
+ "\t 64684598 64684599\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 64684598, 'T.END': 64684599, 'CG': '1='}\n",
+ "7046556 TO1000#1#chr03 568 569\n",
+ "\t 64684598 64684599\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 64684598, 'T.END': 64684599, 'CG': '1='}\n",
+ "7046557 TO1000#1#chr03 569 824\n",
+ "\t 64684600 64684855\n",
+ "{'Q.START': 568, 'Q.END': 569, 'T.START': 64684598, 'T.END': 64684599, 'CG': '1='}\n",
+ "7046558 TO1000#1#chr03 824 826\n",
+ "\t 64684856 64684858\n",
+ "{'Q.START': 569, 'Q.END': 824, 'T.START': 64684600, 'T.END': 64684855, 'CG': '255='}\n",
+ "7046559 TO1000#1#chr03 826 858\n",
+ "\t 64684859 64684891\n",
+ "{'Q.START': 824, 'Q.END': 826, 'T.START': 64684856, 'T.END': 64684858, 'CG': '2='}\n",
+ "7046560 TO1000#1#chr03 858 859\n",
+ "\t 64684892 64684893\n",
+ "{'Q.START': 826, 'Q.END': 858, 'T.START': 64684859, 'T.END': 64684891, 'CG': '32='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 858, 'Q.END': 859, 'T.START': 64684892, 'T.END': 64684893, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046561 TO1000#1#chr03 868 869\n",
+ "\t 64684894 64684895\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046562 TO1000#1#chr03 869 913\n",
+ "\t 64684896 64684940\n",
+ "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
+ "7046564 TO1000#1#chr03 913 919\n",
+ "\t 64684941 64684947\n",
+ "{'Q.START': 869, 'Q.END': 913, 'T.START': 64684896, 'T.END': 64684940, 'CG': '44='}\n",
+ "7046565 TO1000#1#chr03 919 978\n",
+ "\t 64684948 64685007\n",
+ "{'Q.START': 913, 'Q.END': 919, 'T.START': 64684941, 'T.END': 64684947, 'CG': '6='}\n",
+ "7046567 TO1000#1#chr03 978 979\n",
+ "\t 64685008 64685009\n",
+ "{'Q.START': 919, 'Q.END': 978, 'T.START': 64684948, 'T.END': 64685007, 'CG': '59='}\n",
+ "7046568 TO1000#1#chr03 979 1038\n",
+ "\t 64685010 64685069\n",
+ "{'Q.START': 978, 'Q.END': 979, 'T.START': 64685008, 'T.END': 64685009, 'CG': '1='}\n",
+ "7046570 TO1000#1#chr03 1038 1045\n",
+ "\t 64685070 64685077\n",
+ "{'Q.START': 979, 'Q.END': 1038, 'T.START': 64685010, 'T.END': 64685069, 'CG': '59='}\n",
+ "7046571 TO1000#1#chr03 1045 1046\n",
+ "\t 64685078 64685079\n",
+ "{'Q.START': 1038, 'Q.END': 1045, 'T.START': 64685070, 'T.END': 64685077, 'CG': '7='}\n",
+ "7046573 TO1000#1#chr03 1046 1080\n",
+ "\t 64685080 64685114\n",
+ "{'Q.START': 1045, 'Q.END': 1046, 'T.START': 64685078, 'T.END': 64685079, 'CG': '1='}\n",
+ "7046574 TO1000#1#chr03 1080 1081\n",
+ "\t 64685115 64685116\n",
+ "{'Q.START': 1046, 'Q.END': 1080, 'T.START': 64685080, 'T.END': 64685114, 'CG': '34='}\n",
+ "7046576 TO1000#1#chr03 1081 1107\n",
+ "\t 64685117 64685143\n",
+ "{'Q.START': 1080, 'Q.END': 1081, 'T.START': 64685115, 'T.END': 64685116, 'CG': '1='}\n",
+ "7046577 TO1000#1#chr03 1107 1108\n",
+ "\t 64685144 64685145\n",
+ "{'Q.START': 1081, 'Q.END': 1107, 'T.START': 64685117, 'T.END': 64685143, 'CG': '26='}\n",
+ "7046579 TO1000#1#chr03 1108 1183\n",
+ "\t 64685146 64685221\n",
+ "{'Q.START': 1107, 'Q.END': 1108, 'T.START': 64685144, 'T.END': 64685145, 'CG': '1='}\n",
+ "7046581 TO1000#1#chr03 1183 1186\n",
+ "\t 64685222 64685225\n",
+ "{'Q.START': 1108, 'Q.END': 1183, 'T.START': 64685146, 'T.END': 64685221, 'CG': '75='}\n",
+ "7046583 TO1000#1#chr03 1186 1224\n",
+ "\t 64685226 64685264\n",
+ "{'Q.START': 1183, 'Q.END': 1186, 'T.START': 64685222, 'T.END': 64685225, 'CG': '3='}\n",
+ "7046584 TO1000#1#chr03 1224 1257\n",
+ "\t 64685265 64685298\n",
+ "{'Q.START': 1186, 'Q.END': 1224, 'T.START': 64685226, 'T.END': 64685264, 'CG': '38='}\n",
+ "7046586 TO1000#1#chr03 1257 1289\n",
+ "\t 64685299 64685331\n",
+ "{'Q.START': 1224, 'Q.END': 1257, 'T.START': 64685265, 'T.END': 64685298, 'CG': '33='}\n",
+ "7046587 TO1000#1#chr03 1289 1311\n",
+ "\t 64685332 64685354\n",
+ "{'Q.START': 1257, 'Q.END': 1289, 'T.START': 64685299, 'T.END': 64685331, 'CG': '32='}\n",
+ "7046589 TO1000#1#chr03 1311 1359\n",
+ "\t 64685355 64685403\n",
+ "{'Q.START': 1289, 'Q.END': 1311, 'T.START': 64685332, 'T.END': 64685354, 'CG': '22='}\n",
+ "7046590 TO1000#1#chr03 1359 1382\n",
+ "\t 64685404 64685427\n",
+ "{'Q.START': 1311, 'Q.END': 1359, 'T.START': 64685355, 'T.END': 64685403, 'CG': '48='}\n",
+ "7046592 TO1000#1#chr03 1382 1434\n",
+ "\t 64685428 64685480\n",
+ "{'Q.START': 1359, 'Q.END': 1382, 'T.START': 64685404, 'T.END': 64685427, 'CG': '23='}\n",
+ "7046593 TO1000#1#chr03 1434 1451\n",
+ "\t 64685481 64685498\n",
+ "{'Q.START': 1382, 'Q.END': 1434, 'T.START': 64685428, 'T.END': 64685480, 'CG': '52='}\n",
+ "7046594 TO1000#1#chr03 1451 1531\n",
+ "\t 64685499 64685579\n",
+ "{'Q.START': 1434, 'Q.END': 1451, 'T.START': 64685481, 'T.END': 64685498, 'CG': '17='}\n",
+ "7046596 TO1000#1#chr03 1531 1532\n",
+ "\t 64685580 64685581\n",
+ "{'Q.START': 1451, 'Q.END': 1531, 'T.START': 64685499, 'T.END': 64685579, 'CG': '80='}\n",
+ "7046597 TO1000#1#chr03 1532 1543\n",
+ "\t 64685582 64685593\n",
+ "{'Q.START': 1531, 'Q.END': 1532, 'T.START': 64685580, 'T.END': 64685581, 'CG': '1='}\n",
+ "7046599 TO1000#1#chr03 1543 1544\n",
+ "\t 64685594 64685595\n",
+ "{'Q.START': 1532, 'Q.END': 1543, 'T.START': 64685582, 'T.END': 64685593, 'CG': '11='}\n",
+ "7046600 TO1000#1#chr03 1544 1572\n",
+ "\t 64685596 64685624\n",
+ "{'Q.START': 1543, 'Q.END': 1544, 'T.START': 64685594, 'T.END': 64685595, 'CG': '1='}\n",
+ "7046601 TO1000#1#chr03 1572 1573\n",
+ "\t 64685625 64685626\n",
+ "{'Q.START': 1544, 'Q.END': 1572, 'T.START': 64685596, 'T.END': 64685624, 'CG': '28='}\n",
+ "7046603 TO1000#1#chr03 1573 1587\n",
+ "\t 64685627 64685641\n",
+ "{'Q.START': 1572, 'Q.END': 1573, 'T.START': 64685625, 'T.END': 64685626, 'CG': '1='}\n",
+ "7046604 TO1000#1#chr03 1587 1588\n",
+ "\t 64685642 64685643\n",
+ "{'Q.START': 1573, 'Q.END': 1587, 'T.START': 64685627, 'T.END': 64685641, 'CG': '14='}\n",
+ "7046606 TO1000#1#chr03 1588 1616\n",
+ "\t 64685644 64685672\n",
+ "{'Q.START': 1587, 'Q.END': 1588, 'T.START': 64685642, 'T.END': 64685643, 'CG': '1='}\n",
+ "7046608 TO1000#1#chr03 1616 1617\n",
+ "\t 64685673 64685674\n",
+ "{'Q.START': 1588, 'Q.END': 1616, 'T.START': 64685644, 'T.END': 64685672, 'CG': '28='}\n",
+ "7046609 TO1000#1#chr03 1617 1646\n",
+ "\t 64685675 64685704\n",
+ "{'Q.START': 1616, 'Q.END': 1617, 'T.START': 64685673, 'T.END': 64685674, 'CG': '1='}\n",
+ "7046621 TO1000#1#chr03 1646 1661\n",
+ "\t 64685705 64685720\n",
+ "{'Q.START': 1617, 'Q.END': 1646, 'T.START': 64685675, 'T.END': 64685704, 'CG': '29='}\n",
+ "7046622 TO1000#1#chr03 1661 1673\n",
+ "\t 64685721 64685733\n",
+ "{'Q.START': 1646, 'Q.END': 1661, 'T.START': 64685705, 'T.END': 64685720, 'CG': '15='}\n",
+ "7046624 TO1000#1#chr03 1673 1674\n",
+ "\t 64685734 64685735\n",
+ "{'Q.START': 1661, 'Q.END': 1673, 'T.START': 64685721, 'T.END': 64685733, 'CG': '12='}\n",
+ "7046625 TO1000#1#chr03 1674 1726\n",
+ "\t 64685736 64685788\n",
+ "{'Q.START': 1673, 'Q.END': 1674, 'T.START': 64685734, 'T.END': 64685735, 'CG': '1='}\n",
+ "7046626 TO1000#1#chr03 1726 1727\n",
+ "\t 64685789 64685790\n",
+ "{'Q.START': 1674, 'Q.END': 1726, 'T.START': 64685736, 'T.END': 64685788, 'CG': '52='}\n",
+ "7046628 TO1000#1#chr03 1727 1762\n",
+ "\t 64685791 64685826\n",
+ "{'Q.START': 1726, 'Q.END': 1727, 'T.START': 64685789, 'T.END': 64685790, 'CG': '1='}\n",
+ "7046631 TO1000#1#chr03 1766 1767\n",
+ "\t 64685827 64685828\n",
+ "{'Q.START': 1727, 'Q.END': 1762, 'T.START': 64685791, 'T.END': 64685826, 'CG': '35='}\n",
+ "7046673 TO1000#1#chr03 1765 1766\n",
+ "\t 64685829 64685830\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 64685827, 'T.END': 64685828, 'CG': '1='}\n",
+ "7046631 TO1000#1#chr03 1766 1767\n",
+ "\t 64685827 64685828\n",
+ "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 64685829, 'T.END': 64685830, 'CG': '1='}\n",
+ "7046673 TO1000#1#chr03 1765 1766\n",
+ "\t 64685829 64685830\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 64685827, 'T.END': 64685828, 'CG': '1='}\n",
+ "7046631 TO1000#1#chr03 1766 1767\n",
+ "\t 64685827 64685828\n",
+ "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 64685829, 'T.END': 64685830, 'CG': '1='}\n",
+ "7046632 TO1000#1#chr03 1767 1824\n",
+ "\t 64685831 64685888\n",
+ "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 64685827, 'T.END': 64685828, 'CG': '1='}\n",
+ "7046634 TO1000#1#chr03 1824 1825\n",
+ "\t 64685889 64685890\n",
+ "{'Q.START': 1767, 'Q.END': 1824, 'T.START': 64685831, 'T.END': 64685888, 'CG': '57='}\n",
+ "7046635 TO1000#1#chr03 1825 1975\n",
+ "\t 64685891 64686041\n",
+ "{'Q.START': 1824, 'Q.END': 1825, 'T.START': 64685889, 'T.END': 64685890, 'CG': '1='}\n",
+ "7046637 TO1000#1#chr03 1975 1976\n",
+ "\t 64686042 64686043\n",
+ "{'Q.START': 1825, 'Q.END': 1975, 'T.START': 64685891, 'T.END': 64686041, 'CG': '150='}\n",
+ "7046638 TO1000#1#chr03 1976 2015\n",
+ "\t 64686044 64686083\n",
+ "{'Q.START': 1975, 'Q.END': 1976, 'T.START': 64686042, 'T.END': 64686043, 'CG': '1='}\n",
+ "7046639 TO1000#1#chr03 2015 2016\n",
+ "\t 64686084 64686085\n",
+ "{'Q.START': 1976, 'Q.END': 2015, 'T.START': 64686044, 'T.END': 64686083, 'CG': '39='}\n",
+ "7046641 TO1000#1#chr03 2016 2047\n",
+ "\t 64686086 64686117\n",
+ "{'Q.START': 2015, 'Q.END': 2016, 'T.START': 64686084, 'T.END': 64686085, 'CG': '1='}\n",
+ "7046644 TO1000#1#chr03 2047 2055\n",
+ "\t 64686118 64686126\n",
+ "{'Q.START': 2016, 'Q.END': 2047, 'T.START': 64686086, 'T.END': 64686117, 'CG': '31='}\n",
+ "7046646 TO1000#1#chr03 2055 2056\n",
+ "\t 64686127 64686128\n",
+ "{'Q.START': 2047, 'Q.END': 2055, 'T.START': 64686118, 'T.END': 64686126, 'CG': '8='}\n",
+ "7046647 TO1000#1#chr03 2056 2120\n",
+ "\t 64686129 64686193\n",
+ "{'Q.START': 2055, 'Q.END': 2056, 'T.START': 64686127, 'T.END': 64686128, 'CG': '1='}\n",
+ "7046649 TO1000#1#chr03 2120 2121\n",
+ "\t 64686194 64686195\n",
+ "{'Q.START': 2056, 'Q.END': 2120, 'T.START': 64686129, 'T.END': 64686193, 'CG': '64='}\n",
+ "7046650 TO1000#1#chr03 2121 2157\n",
+ "\t 64686196 64686232\n",
+ "{'Q.START': 2120, 'Q.END': 2121, 'T.START': 64686194, 'T.END': 64686195, 'CG': '1='}\n",
+ "7046652 TO1000#1#chr03 2157 2158\n",
+ "\t 64686233 64686234\n",
+ "{'Q.START': 2121, 'Q.END': 2157, 'T.START': 64686196, 'T.END': 64686232, 'CG': '36='}\n",
+ "7046653 TO1000#1#chr03 2158 2170\n",
+ "\t 64686235 64686247\n",
+ "{'Q.START': 2157, 'Q.END': 2158, 'T.START': 64686233, 'T.END': 64686234, 'CG': '1='}\n",
+ "7046654 TO1000#1#chr03 2170 2171\n",
+ "\t 64686248 64686249\n",
+ "{'Q.START': 2158, 'Q.END': 2170, 'T.START': 64686235, 'T.END': 64686247, 'CG': '12='}\n",
+ "7046656 TO1000#1#chr03 2171 2205\n",
+ "\t 64686250 64686284\n",
+ "{'Q.START': 2170, 'Q.END': 2171, 'T.START': 64686248, 'T.END': 64686249, 'CG': '1='}\n",
+ "7046657 TO1000#1#chr03 2205 2206\n",
+ "\t 64686285 64686286\n",
+ "{'Q.START': 2171, 'Q.END': 2205, 'T.START': 64686250, 'T.END': 64686284, 'CG': '34='}\n",
+ "7046659 TO1000#1#chr03 2206 2344\n",
+ "\t 64686287 64686425\n",
+ "{'Q.START': 2205, 'Q.END': 2206, 'T.START': 64686285, 'T.END': 64686286, 'CG': '1='}\n",
+ "7046660 TO1000#1#chr03 2344 2345\n",
+ "\t 64686426 64686427\n",
+ "{'Q.START': 2206, 'Q.END': 2344, 'T.START': 64686287, 'T.END': 64686425, 'CG': '138='}\n",
+ "7046662 TO1000#1#chr03 2345 2364\n",
+ "\t 64686428 64686447\n",
+ "{'Q.START': 2344, 'Q.END': 2345, 'T.START': 64686426, 'T.END': 64686427, 'CG': '1='}\n",
+ "7046663 TO1000#1#chr03 2364 2383\n",
+ "\t 64686448 64686467\n",
+ "{'Q.START': 2345, 'Q.END': 2364, 'T.START': 64686428, 'T.END': 64686447, 'CG': '19='}\n",
+ "7046665 TO1000#1#chr03 2383 2408\n",
+ "\t 64686468 64686493\n",
+ "{'Q.START': 2364, 'Q.END': 2383, 'T.START': 64686448, 'T.END': 64686467, 'CG': '19='}\n",
+ "7046667 TO1000#1#chr03 2408 2409\n",
+ "\t 64686494 64686495\n",
+ "{'Q.START': 2383, 'Q.END': 2408, 'T.START': 64686468, 'T.END': 64686493, 'CG': '25='}\n",
+ "7046668 TO1000#1#chr03 2409 2441\n",
+ "\t 64686496 64686528\n",
+ "{'Q.START': 2408, 'Q.END': 2409, 'T.START': 64686494, 'T.END': 64686495, 'CG': '1='}\n",
+ "7046670 TO1000#1#chr03 2441 2442\n",
+ "\t 64686529 64686530\n",
+ "{'Q.START': 2409, 'Q.END': 2441, 'T.START': 64686496, 'T.END': 64686528, 'CG': '32='}\n",
+ "7046671 TO1000#1#chr03 2442 2580\n",
+ "\t 64686531 64686669\n",
+ "{'Q.START': 2441, 'Q.END': 2442, 'T.START': 64686529, 'T.END': 64686530, 'CG': '1='}\n",
+ "7046674 TO1000#1#chr03 2582 2583\n",
+ "\t 64686670 64686671\n",
+ "{'Q.START': 2442, 'Q.END': 2580, 'T.START': 64686531, 'T.END': 64686669, 'CG': '138='}\n",
+ "7046675 TO1000#1#chr03 2583 2584\n",
+ "\t 64686672 64686673\n",
+ "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 64686670, 'T.END': 64686671, 'CG': '1='}\n",
+ "7046674 TO1000#1#chr03 2582 2583\n",
+ "\t 64686670 64686671\n",
+ "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 64686672, 'T.END': 64686673, 'CG': '1='}\n",
+ "7046675 TO1000#1#chr03 2583 2584\n",
+ "\t 64686672 64686673\n",
+ "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 64686670, 'T.END': 64686671, 'CG': '1='}\n",
+ "7046676 TO1000#1#chr03 2584 2764\n",
+ "\t 64686674 64686854\n",
+ "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 64686672, 'T.END': 64686673, 'CG': '1='}\n",
+ "7046678 TO1000#1#chr03 2764 2765\n",
+ "\t 64686855 64686856\n",
+ "{'Q.START': 2584, 'Q.END': 2764, 'T.START': 64686674, 'T.END': 64686854, 'CG': '180='}\n",
+ "7046679 TO1000#1#chr03 2765 2797\n",
+ "\t 64686857 64686889\n",
+ "{'Q.START': 2764, 'Q.END': 2765, 'T.START': 64686855, 'T.END': 64686856, 'CG': '1='}\n",
+ "7046680 TO1000#1#chr03 2797 2798\n",
+ "\t 64686890 64686891\n",
+ "{'Q.START': 2765, 'Q.END': 2797, 'T.START': 64686857, 'T.END': 64686889, 'CG': '32='}\n",
+ "7046682 TO1000#1#chr03 2798 2878\n",
+ "\t 64686892 64686972\n",
+ "{'Q.START': 2797, 'Q.END': 2798, 'T.START': 64686890, 'T.END': 64686891, 'CG': '1='}\n",
+ "7046684 TO1000#1#chr03 2878 2879\n",
+ "\t 64686973 64686974\n",
+ "{'Q.START': 2798, 'Q.END': 2878, 'T.START': 64686892, 'T.END': 64686972, 'CG': '80='}\n",
+ "7046685 TO1000#1#chr03 2879 2951\n",
+ "\t 64686975 64687047\n",
+ "{'Q.START': 2878, 'Q.END': 2879, 'T.START': 64686973, 'T.END': 64686974, 'CG': '1='}\n",
+ "7046686 TO1000#1#chr03 2951 2952\n",
+ "\t 64687048 64687049\n",
+ "{'Q.START': 2879, 'Q.END': 2951, 'T.START': 64686975, 'T.END': 64687047, 'CG': '72='}\n",
+ "7046688 TO1000#1#chr03 2952 3002\n",
+ "\t 64687050 64687100\n",
+ "{'Q.START': 2951, 'Q.END': 2952, 'T.START': 64687048, 'T.END': 64687049, 'CG': '1='}\n",
+ "7046690 TO1000#1#chr03 3002 3077\n",
+ "\t 64687101 64687176\n",
+ "{'Q.START': 2952, 'Q.END': 3002, 'T.START': 64687050, 'T.END': 64687100, 'CG': '50='}\n",
+ "7046692 TO1000#1#chr03 3077 3078\n",
+ "\t 64687177 64687178\n",
+ "{'Q.START': 3002, 'Q.END': 3077, 'T.START': 64687101, 'T.END': 64687176, 'CG': '75='}\n",
+ "7046693 TO1000#1#chr03 3078 3093\n",
+ "\t 64687179 64687194\n",
+ "{'Q.START': 3077, 'Q.END': 3078, 'T.START': 64687177, 'T.END': 64687178, 'CG': '1='}\n",
+ "7046695 TO1000#1#chr03 3093 3094\n",
+ "\t 64687195 64687196\n",
+ "{'Q.START': 3078, 'Q.END': 3093, 'T.START': 64687179, 'T.END': 64687194, 'CG': '15='}\n",
+ "7046696 TO1000#1#chr03 3094 3097\n",
+ "\t 64687197 64687200\n",
+ "{'Q.START': 3093, 'Q.END': 3094, 'T.START': 64687195, 'T.END': 64687196, 'CG': '1='}\n",
+ "7046698 TO1000#1#chr03 3097 3140\n",
+ "\t 64687201 64687244\n",
+ "{'Q.START': 3094, 'Q.END': 3097, 'T.START': 64687197, 'T.END': 64687200, 'CG': '3='}\n",
+ "7046700 TO1000#1#chr03 3140 3210\n",
+ "\t 64687245 64687315\n",
+ "{'Q.START': 3097, 'Q.END': 3140, 'T.START': 64687201, 'T.END': 64687244, 'CG': '43='}\n",
+ "7046702 TO1000#1#chr03 3210 3211\n",
+ "\t 64687316 64687317\n",
+ "{'Q.START': 3140, 'Q.END': 3210, 'T.START': 64687245, 'T.END': 64687315, 'CG': '70='}\n",
+ "7046703 TO1000#1#chr03 3211 3229\n",
+ "\t 64687318 64687336\n",
+ "{'Q.START': 3210, 'Q.END': 3211, 'T.START': 64687316, 'T.END': 64687317, 'CG': '1='}\n",
+ "7046704 TO1000#1#chr03 3229 3230\n",
+ "\t 64687337 64687338\n",
+ "{'Q.START': 3211, 'Q.END': 3229, 'T.START': 64687318, 'T.END': 64687336, 'CG': '18='}\n",
+ "7046706 TO1000#1#chr03 3230 3276\n",
+ "\t 64687339 64687385\n",
+ "{'Q.START': 3229, 'Q.END': 3230, 'T.START': 64687337, 'T.END': 64687338, 'CG': '1='}\n",
+ "7046707 TO1000#1#chr03 3276 3277\n",
+ "\t 64687386 64687387\n",
+ "{'Q.START': 3230, 'Q.END': 3276, 'T.START': 64687339, 'T.END': 64687385, 'CG': '46='}\n",
+ "7046709 TO1000#1#chr03 3277 3315\n",
+ "\t 64687388 64687426\n",
+ "{'Q.START': 3276, 'Q.END': 3277, 'T.START': 64687386, 'T.END': 64687387, 'CG': '1='}\n",
+ "7046710 TO1000#1#chr03 3315 3316\n",
+ "\t 64687427 64687428\n",
+ "{'Q.START': 3277, 'Q.END': 3315, 'T.START': 64687388, 'T.END': 64687426, 'CG': '38='}\n",
+ "7046712 TO1000#1#chr03 3316 3322\n",
+ "\t 64687429 64687435\n",
+ "{'Q.START': 3315, 'Q.END': 3316, 'T.START': 64687427, 'T.END': 64687428, 'CG': '1='}\n",
+ "7046713 TO1000#1#chr03 3322 3323\n",
+ "\t 64687436 64687437\n",
+ "{'Q.START': 3316, 'Q.END': 3322, 'T.START': 64687429, 'T.END': 64687435, 'CG': '6='}\n",
+ "7046715 TO1000#1#chr03 3323 3348\n",
+ "\t 64687438 64687463\n",
+ "{'Q.START': 3322, 'Q.END': 3323, 'T.START': 64687436, 'T.END': 64687437, 'CG': '1='}\n",
+ "7046718 TO1000#1#chr03 3352 3353\n",
+ "\t 64687464 64687465\n",
+ "{'Q.START': 3323, 'Q.END': 3348, 'T.START': 64687438, 'T.END': 64687463, 'CG': '25='}\n",
+ "7046717 TO1000#1#chr03 3351 3352\n",
+ "\t 64687466 64687467\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 64687464, 'T.END': 64687465, 'CG': '1='}\n",
+ "7046718 TO1000#1#chr03 3352 3353\n",
+ "\t 64687464 64687465\n",
+ "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 64687466, 'T.END': 64687467, 'CG': '1='}\n",
+ "7046717 TO1000#1#chr03 3351 3352\n",
+ "\t 64687466 64687467\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 64687464, 'T.END': 64687465, 'CG': '1='}\n",
+ "7046718 TO1000#1#chr03 3352 3353\n",
+ "\t 64687464 64687465\n",
+ "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 64687466, 'T.END': 64687467, 'CG': '1='}\n",
+ "7046720 TO1000#1#chr03 3353 3354\n",
+ "\t 64687468 64687469\n",
+ "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 64687464, 'T.END': 64687465, 'CG': '1='}\n",
+ "7046722 TO1000#1#chr03 3354 3356\n",
+ "\t 64687470 64687472\n",
+ "{'Q.START': 3353, 'Q.END': 3354, 'T.START': 64687468, 'T.END': 64687469, 'CG': '1='}\n",
+ "7046724 TO1000#1#chr03 3356 3357\n",
+ "\t 64687473 64687474\n",
+ "{'Q.START': 3354, 'Q.END': 3356, 'T.START': 64687470, 'T.END': 64687472, 'CG': '2='}\n",
+ "7046725 TO1000#1#chr03 3357 3489\n",
+ "\t 64687475 64687607\n",
+ "{'Q.START': 3356, 'Q.END': 3357, 'T.START': 64687473, 'T.END': 64687474, 'CG': '1='}\n",
+ "7046727 TO1000#1#chr03 3489 3490\n",
+ "\t 64687608 64687609\n",
+ "{'Q.START': 3357, 'Q.END': 3489, 'T.START': 64687475, 'T.END': 64687607, 'CG': '132='}\n",
+ "7046728 TO1000#1#chr03 3490 3642\n",
+ "\t 64687610 64687762\n",
+ "{'Q.START': 3489, 'Q.END': 3490, 'T.START': 64687608, 'T.END': 64687609, 'CG': '1='}\n",
+ "7046729 TO1000#1#chr03 3642 3644\n",
+ "\t 64687763 64687765\n",
+ "{'Q.START': 3490, 'Q.END': 3642, 'T.START': 64687610, 'T.END': 64687762, 'CG': '152='}\n",
+ "7046730 TO1000#1#chr03 3644 3685\n",
+ "\t 64687766 64687807\n",
+ "{'Q.START': 3642, 'Q.END': 3644, 'T.START': 64687763, 'T.END': 64687765, 'CG': '2='}\n",
+ "7046731 TO1000#1#chr03 3685 3687\n",
+ "\t 64687808 64687810\n",
+ "{'Q.START': 3644, 'Q.END': 3685, 'T.START': 64687766, 'T.END': 64687807, 'CG': '41='}\n",
+ "7046733 TO1000#1#chr03 3687 3693\n",
+ "\t 64687811 64687817\n",
+ "{'Q.START': 3685, 'Q.END': 3687, 'T.START': 64687808, 'T.END': 64687810, 'CG': '2='}\n",
+ "7046735 TO1000#1#chr03 3693 3694\n",
+ "\t 64687818 64687819\n",
+ "{'Q.START': 3687, 'Q.END': 3693, 'T.START': 64687811, 'T.END': 64687817, 'CG': '6='}\n",
+ "7046736 TO1000#1#chr03 3694 3708\n",
+ "\t 64687820 64687834\n",
+ "{'Q.START': 3693, 'Q.END': 3694, 'T.START': 64687818, 'T.END': 64687819, 'CG': '1='}\n",
+ "7046738 TO1000#1#chr03 3720 3721\n",
+ "\t 64687835 64687836\n",
+ "{'Q.START': 3694, 'Q.END': 3708, 'T.START': 64687820, 'T.END': 64687834, 'CG': '14='}\n",
+ "7046739 TO1000#1#chr03 3721 3722\n",
+ "\t 64687837 64687838\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 64687835, 'T.END': 64687836, 'CG': '1='}\n",
+ "7046740 TO1000#1#chr03 3716 3720\n",
+ "\t 64687839 64687843\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 64687837, 'T.END': 64687838, 'CG': '1='}\n",
+ "7046738 TO1000#1#chr03 3720 3721\n",
+ "\t 64687835 64687836\n",
+ "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 64687839, 'T.END': 64687843, 'CG': '4='}\n",
+ "7046739 TO1000#1#chr03 3721 3722\n",
+ "\t 64687837 64687838\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 64687835, 'T.END': 64687836, 'CG': '1='}\n",
+ "7046740 TO1000#1#chr03 3716 3720\n",
+ "\t 64687839 64687843\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 64687837, 'T.END': 64687838, 'CG': '1='}\n",
+ "7046738 TO1000#1#chr03 3720 3721\n",
+ "\t 64687835 64687836\n",
+ "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 64687839, 'T.END': 64687843, 'CG': '4='}\n",
+ "7046739 TO1000#1#chr03 3721 3722\n",
+ "\t 64687837 64687838\n",
+ "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 64687835, 'T.END': 64687836, 'CG': '1='}\n",
+ "7046741 TO1000#1#chr03 3722 3735\n",
+ "\t 64687844 64687877\n",
+ "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 64687837, 'T.END': 64687838, 'CG': '1='}\n",
+ "ALN_2\n",
+ "7594382 D134#1#chr03 0 1\n",
+ "\tIn path\n",
+ "\t 70220037 70220038\n",
+ "skipped\n",
+ "\n",
+ "7594369 D134#1#chr03 32 33\n",
+ "\tIn path\n",
+ "\t 70219216 70219217\n",
+ "{'Q.START': 0, 'Q.END': 1, 'T.START': 70220037, 'T.END': 70220038, 'CG': '1='}\n",
+ "7594371 D134#1#chr03 15 16\n",
+ "\tIn path\n",
+ "\t 70221163 70221164\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 15, 'Q.END': 16, 'T.START': 70221163, 'T.END': 70221164, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594356 D134#1#chr03 66 67\n",
+ "\tIn path\n",
+ "\t 70219570 70219571\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594626 D134#1#chr03 10 11\n",
+ "\tIn path\n",
+ "\t 70219214 70219215\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594011 D134#1#chr03 11 12\n",
+ "\tIn path\n",
+ "\t 70219995 70219996\n",
+ "{'Q.START': 10, 'Q.END': 11, 'T.START': 70219214, 'T.END': 70219215, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 11, 'Q.END': 12, 'T.START': 70219995, 'T.END': 70219996, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594369 D134#1#chr03 32 33\n",
+ "\tIn path\n",
+ "\t 70219216 70219217\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594371 D134#1#chr03 15 16\n",
+ "\tIn path\n",
+ "\t 70221163 70221164\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 15, 'Q.END': 16, 'T.START': 70221163, 'T.END': 70221164, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594241 D134#1#chr03 20 21\n",
+ "\tIn path\n",
+ "\t 70219220 70219221\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594248 D134#1#chr03 21 22\n",
+ "\tNot in path\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 20, 'Q.END': 21, 'T.START': 70219220, 'T.END': 70219221, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594315 D134#1#chr03 53 54\n",
+ "\tIn path\n",
+ "\t 70219857 70219858\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
+ "7594330 D134#1#chr03 26 27\n",
+ "\tNot in path\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594315 D134#1#chr03 53 54\n",
+ "\tIn path\n",
+ "\t 70219857 70219858\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594369 D134#1#chr03 32 33\n",
+ "\tIn path\n",
+ "\t 70219216 70219217\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
+ "7594026 D134#1#chr03 37 38\n",
+ "\tIn path\n",
+ "\t 70220249 70220250\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 37, 'Q.END': 38, 'T.START': 70220249, 'T.END': 70220250, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594026 D134#1#chr03 37 38\n",
+ "\tIn path\n",
+ "\t 70220249 70220250\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 37, 'Q.END': 38, 'T.START': 70220249, 'T.END': 70220250, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594315 D134#1#chr03 53 54\n",
+ "\tIn path\n",
+ "\t 70219857 70219858\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
+ "7594311 D134#1#chr03 55 56\n",
+ "\tIn path\n",
+ "\t 70219351 70219352\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
+ "7594021 D134#1#chr03 57 58\n",
+ "\tIn path\n",
+ "\t 70219218 70219219\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
+ "7594286 D134#1#chr03 59 60\n",
+ "\tIn path\n",
+ "\t 70219349 70219350\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
+ "7594356 D134#1#chr03 66 67\n",
+ "\tIn path\n",
+ "\t 70219570 70219571\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594356 D134#1#chr03 66 67\n",
+ "\tIn path\n",
+ "\t 70219570 70219571\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
+ "7594375 D134#1#chr03 68 69\n",
+ "\tIn path\n",
+ "\t 70221598 70221599\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594374 D134#1#chr03 69 70\n",
+ "\tIn path\n",
+ "\t 70219092 70219093\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
+ "7594350 D134#1#chr03 70 71\n",
+ "\tIn path\n",
+ "\t 70219226 70219227\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
+ "7594264 D134#1#chr03 71 72\n",
+ "\tIn path\n",
+ "\t 70219228 70219229\n",
+ "{'Q.START': 70, 'Q.END': 71, 'T.START': 70219226, 'T.END': 70219227, 'CG': '1='}\n",
+ "7594207 D134#1#chr03 72 73\n",
+ "\tIn path\n",
+ "\t 70219230 70219231\n",
+ "{'Q.START': 71, 'Q.END': 72, 'T.START': 70219228, 'T.END': 70219229, 'CG': '1='}\n",
+ "7594225 D134#1#chr03 73 74\n",
+ "\tIn path\n",
+ "\t 70219232 70219233\n",
+ "{'Q.START': 72, 'Q.END': 73, 'T.START': 70219230, 'T.END': 70219231, 'CG': '1='}\n",
+ "7594227 D134#1#chr03 74 75\n",
+ "\tIn path\n",
+ "\t 70220150 70220151\n",
+ "{'Q.START': 73, 'Q.END': 74, 'T.START': 70219232, 'T.END': 70219233, 'CG': '1='}\n",
+ "7594120 D134#1#chr03 75 76\n",
+ "\tIn path\n",
+ "\t 70219236 70219237\n",
+ "{'Q.START': 74, 'Q.END': 75, 'T.START': 70220150, 'T.END': 70220151, 'CG': '1='}\n",
+ "7594132 D134#1#chr03 76 77\n",
+ "\tIn path\n",
+ "\t 70219777 70219778\n",
+ "{'Q.START': 75, 'Q.END': 76, 'T.START': 70219236, 'T.END': 70219237, 'CG': '1='}\n",
+ "7594165 D134#1#chr03 77 78\n",
+ "\tIn path\n",
+ "\t 70219240 70219241\n",
+ "{'Q.START': 76, 'Q.END': 77, 'T.START': 70219777, 'T.END': 70219778, 'CG': '1='}\n",
+ "7594172 D134#1#chr03 78 3735\n",
+ "\tNot in path\n",
+ "ALN_2\n",
+ "7594382 TO1000#1#chr03 0 1\n",
+ "\t 61731222 61731223\n",
+ "skipped\n",
+ "\n",
+ "7594369 TO1000#1#chr03 32 33\n",
+ "\t 61731060 61731061\n",
+ "{'Q.START': 0, 'Q.END': 1, 'T.START': 61731222, 'T.END': 61731223, 'CG': '1='}\n",
+ "7594371 TO1000#1#chr03 15 16\n",
+ "\tNot in path\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 61731060, 'T.END': 61731061, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594356 TO1000#1#chr03 66 67\n",
+ "\t 61731519 61731520\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 61731519, 'T.END': 61731520, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594375 TO1000#1#chr03 68 69\n",
+ "\t 61733612 61733613\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594626 TO1000#1#chr03 10 11\n",
+ "\t 61731056 61731057\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
+ "7594011 TO1000#1#chr03 11 12\n",
+ "\t 61733900 61733901\n",
+ "{'Q.START': 10, 'Q.END': 11, 'T.START': 61731056, 'T.END': 61731057, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 11, 'Q.END': 12, 'T.START': 61733900, 'T.END': 61733901, 'CG': '1='}\n",
+ "7594375 TO1000#1#chr03 68 69\n",
+ "\t 61733612 61733613\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594369 TO1000#1#chr03 32 33\n",
+ "\t 61731060 61731061\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
+ "7594371 TO1000#1#chr03 15 16\n",
+ "\tNot in path\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 61731060, 'T.END': 61731061, 'CG': '1='}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594241 TO1000#1#chr03 20 21\n",
+ "\t 61731046 61731047\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594248 TO1000#1#chr03 21 22\n",
+ "\t 61734261 61734262\n",
+ "{'Q.START': 20, 'Q.END': 21, 'T.START': 61731046, 'T.END': 61731047, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 21, 'Q.END': 22, 'T.START': 61734261, 'T.END': 61734262, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594315 TO1000#1#chr03 53 54\n",
+ "\t 61733937 61733938\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 61733937, 'T.END': 61733938, 'CG': '1='}\n",
+ "7594330 TO1000#1#chr03 26 27\n",
+ "\t 61731768 61731769\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 26, 'Q.END': 27, 'T.START': 61731768, 'T.END': 61731769, 'CG': '1='}\n",
+ "7594315 TO1000#1#chr03 53 54\n",
+ "\t 61733937 61733938\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 61733937, 'T.END': 61733938, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594369 TO1000#1#chr03 32 33\n",
+ "\t 61731060 61731061\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 32, 'Q.END': 33, 'T.START': 61731060, 'T.END': 61731061, 'CG': '1='}\n",
+ "7594026 TO1000#1#chr03 37 38\n",
+ "\t 61734267 61734268\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 37, 'Q.END': 38, 'T.START': 61734267, 'T.END': 61734268, 'CG': '1='}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594026 TO1000#1#chr03 37 38\n",
+ "\t 61734267 61734268\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 37, 'Q.END': 38, 'T.START': 61734267, 'T.END': 61734268, 'CG': '1='}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594315 TO1000#1#chr03 53 54\n",
+ "\t 61733937 61733938\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 53, 'Q.END': 54, 'T.START': 61733937, 'T.END': 61733938, 'CG': '1='}\n",
+ "7594311 TO1000#1#chr03 55 56\n",
+ "\t 61731052 61731053\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
+ "7594021 TO1000#1#chr03 57 58\n",
+ "\t 61730922 61730923\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
+ "7594286 TO1000#1#chr03 59 60\n",
+ "\t 61731054 61731055\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
+ "7594356 TO1000#1#chr03 66 67\n",
+ "\t 61731519 61731520\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 61731519, 'T.END': 61731520, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594375 TO1000#1#chr03 68 69\n",
+ "\t 61733612 61733613\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
+ "7594356 TO1000#1#chr03 66 67\n",
+ "\t 61731519 61731520\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 66, 'Q.END': 67, 'T.START': 61731519, 'T.END': 61731520, 'CG': '1='}\n",
+ "7594375 TO1000#1#chr03 68 69\n",
+ "\t 61733612 61733613\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594374 TO1000#1#chr03 69 70\n",
+ "\t 61730920 61730921\n",
+ "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
+ "7594350 TO1000#1#chr03 70 71\n",
+ "\t 61731066 61731067\n",
+ "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
+ "7594264 TO1000#1#chr03 71 72\n",
+ "\t 61731068 61731069\n",
+ "{'Q.START': 70, 'Q.END': 71, 'T.START': 61731066, 'T.END': 61731067, 'CG': '1='}\n",
+ "7594207 TO1000#1#chr03 72 73\n",
+ "\t 61731070 61731071\n",
+ "{'Q.START': 71, 'Q.END': 72, 'T.START': 61731068, 'T.END': 61731069, 'CG': '1='}\n",
+ "7594225 TO1000#1#chr03 73 74\n",
+ "\t 61731072 61731073\n",
+ "{'Q.START': 72, 'Q.END': 73, 'T.START': 61731070, 'T.END': 61731071, 'CG': '1='}\n",
+ "7594227 TO1000#1#chr03 74 75\n",
+ "\tNot in path\n",
+ "7594120 TO1000#1#chr03 75 76\n",
+ "\t 61731076 61731077\n",
+ "{'Q.START': 73, 'Q.END': 74, 'T.START': 61731072, 'T.END': 61731073, 'CG': '1='}\n",
+ "7594132 TO1000#1#chr03 76 77\n",
+ "\t 61733800 61733801\n",
+ "{'Q.START': 75, 'Q.END': 76, 'T.START': 61731076, 'T.END': 61731077, 'CG': '1='}\n",
+ "7594165 TO1000#1#chr03 77 78\n",
+ "\t 61731080 61731081\n",
+ "{'Q.START': 76, 'Q.END': 77, 'T.START': 61733800, 'T.END': 61733801, 'CG': '1='}\n",
+ "7594172 TO1000#1#chr03 78 3735\n",
+ "\tNot in path\n"
+ ]
+ }
+ ],
+ "source": [
+ "ALNS = {}\n",
+ "## Iterating over alignments\n",
+ "for aln_name in aln_dict.keys():\n",
+ " \n",
+ " ## Iterating over paths of the gfa\n",
+ " for path_name in paths.keys():\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(aln_name)\n",
+ " _ = [] # Temporary list holding alignment blocks\n",
+ "\n",
+ " ## Iterating over alignment nodes of the current alignment\n",
+ " for node_id, orient in aln_dict[aln_name][\"PATH.MATCH\"]:\n",
+ "\n",
+ " # Getting node info\n",
+ " n_info = nodes[node_id]\n",
+ " q_start = n_info[aln_name][\"START\"] # Start position on the query\n",
+ " q_end = n_info[aln_name][\"END\"] # End position on the query\n",
+ " _CG = n_info[aln_name][\"CIGAR\"] # Cigar of the alignment on the current node\n",
+ "\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(node_id, path_name, q_start, q_end)\n",
+ "\n",
+ " ## Checking if path is traversing the current node\n",
+ " if path_name in list(n_info.keys()):\n",
+ " if path_name == \"D134#1#chr03\": print(\"\\tIn path\")\n",
+ "\n",
+ " ## Getting start and end position on the target given the orientation of the node in the alignment and the path\n",
+ " if n_info[aln_name][\"STRAND\"] == n_info[path_name][\"STRAND\"] :\n",
+ " t_start = n_info[path_name][\"START\"]+n_info[aln_name][\"S.OFF\"]\n",
+ " t_end = n_info[path_name][\"END\"]+n_info[aln_name][\"E.OFF\"] \n",
+ " else :\n",
+ " t_end = n_info[path_name][\"START\"]+n_info[aln_name][\"S.OFF\"]\n",
+ " t_start = n_info[path_name][\"END\"]+n_info[aln_name][\"E.OFF\"]\n",
+ "\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"\\t\", t_start, t_end)\n",
+ "\n",
+ " \"\"\"\n",
+ " If the latest block t.end and q.end matches with the current node t.start and q.start, \n",
+ " the node should be added to the block. Else, we terminate the block and add the node to a new block\n",
+ " \"\"\"\n",
+ " \n",
+ " # Non empty temporary list of aln and ending of the last block is the same as the start of the new node : \n",
+ " if len(_) and _[-1][\"T.END\"] == t_start and _[-1][\"Q.END\"]+1 == q_start: \n",
+ " tmp_aln[\"Q.END\"] = q_end\n",
+ " tmp_aln[\"T.END\"] = t_end\n",
+ " tmp_aln[\"CG\"] += _CG\n",
+ "# elif len(_) and _[-1][\"T.END\"] == t_start: # Following on the target not on the query (i.e. Insertion)\n",
+ "# tmp_aln[\"T.END\"] = t_end\n",
+ "# tmp_aln[\"CG\"] += f\"{nodes_length[node_id]}I\"\n",
+ "# elif len(_) and _[-1][\"Q.END\"]+1 == q_start: # Following on the query, not on the target (i.e. Deletion)\n",
+ "# tmp_aln[\"Q.END\"] = q_end\n",
+ "# tmp_aln[\"CG\"] += f\"{nodes_length[node_id]}D\"\n",
+ " else : # Else, completely different\n",
+ " try : \n",
+ " _.append(tmp_aln)\n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(tmp_aln)\n",
+ " except : \n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"skipped\\n\")\n",
+ " tmp_aln = {\n",
+ " \"Q.START\": q_start,\n",
+ " \"Q.END\": q_end,\n",
+ " \"T.START\": t_start,\n",
+ " \"T.END\": t_end,\n",
+ " \"CG\": _CG,\n",
+ " }\n",
+ " \n",
+ " else : \n",
+ " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"\\tNot in path\")\n",
+ " # Node is not in the path\n",
+ "\n",
+ " del tmp_aln\n",
+ " \n",
+ " ALNS[(path_name, aln_name)] = _"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 9,
+ "id": "547f03fa-cbd5-42f9-b668-1ca4404795ba",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "[{'Q.START': 0, 'Q.END': 77, 'T.START': 73306158, 'T.END': 73306235, 'CG': '77='}, {'Q.START': 77, 'Q.END': 82, 'T.START': 73306238, 'T.END': 73306243, 'CG': '5='}, {'Q.START': 83, 'Q.END': 138, 'T.START': 73306246, 'T.END': 73306301, 'CG': '55='}, {'Q.START': 139, 'Q.END': 202, 'T.START': 73306302, 'T.END': 73306365, 'CG': '63='}, {'Q.START': 202, 'Q.END': 203, 'T.START': 73306366, 'T.END': 73306367, 'CG': '1='}, {'Q.START': 203, 'Q.END': 379, 'T.START': 73306368, 'T.END': 73306544, 'CG': '176='}, {'Q.START': 379, 'Q.END': 380, 'T.START': 73306545, 'T.END': 73306546, 'CG': '1='}, {'Q.START': 380, 'Q.END': 429, 'T.START': 73306547, 'T.END': 73306596, 'CG': '49='}, {'Q.START': 429, 'Q.END': 430, 'T.START': 73306597, 'T.END': 73306598, 'CG': '1='}, {'Q.START': 430, 'Q.END': 457, 'T.START': 73306599, 'T.END': 73306626, 'CG': '27='}, {'Q.START': 457, 'Q.END': 492, 'T.START': 73306641, 'T.END': 73306676, 'CG': '35='}, {'Q.START': 508, 'Q.END': 564, 'T.START': 73306694, 'T.END': 73306750, 'CG': '56='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 569, 'Q.END': 824, 'T.START': 73306755, 'T.END': 73307010, 'CG': '255='}, {'Q.START': 826, 'Q.END': 858, 'T.START': 73307011, 'T.END': 73307043, 'CG': '32='}, {'Q.START': 858, 'Q.END': 859, 'T.START': 73307044, 'T.END': 73307045, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 869, 'Q.END': 913, 'T.START': 73307048, 'T.END': 73307092, 'CG': '44='}, {'Q.START': 913, 'Q.END': 919, 'T.START': 73307093, 'T.END': 73307099, 'CG': '6='}, {'Q.START': 919, 'Q.END': 978, 'T.START': 73307100, 'T.END': 73307159, 'CG': '59='}, {'Q.START': 978, 'Q.END': 979, 'T.START': 73307160, 'T.END': 73307161, 'CG': '1='}, {'Q.START': 979, 'Q.END': 1038, 'T.START': 73307162, 'T.END': 73307221, 'CG': '59='}, {'Q.START': 1038, 'Q.END': 1045, 'T.START': 73307224, 'T.END': 73307231, 'CG': '7='}, {'Q.START': 1045, 'Q.END': 1046, 'T.START': 73307232, 'T.END': 73307233, 'CG': '1='}, {'Q.START': 1046, 'Q.END': 1080, 'T.START': 73307234, 'T.END': 73307268, 'CG': '34='}, {'Q.START': 1080, 'Q.END': 1081, 'T.START': 73307269, 'T.END': 73307270, 'CG': '1='}, {'Q.START': 1081, 'Q.END': 1107, 'T.START': 73307271, 'T.END': 73307297, 'CG': '26='}, {'Q.START': 1108, 'Q.END': 1183, 'T.START': 73307300, 'T.END': 73307375, 'CG': '75='}, {'Q.START': 1183, 'Q.END': 1186, 'T.START': 73307376, 'T.END': 73307379, 'CG': '3='}, {'Q.START': 1224, 'Q.END': 1257, 'T.START': 73307419, 'T.END': 73307452, 'CG': '33='}, {'Q.START': 1289, 'Q.END': 1311, 'T.START': 73307475, 'T.END': 73307497, 'CG': '22='}, {'Q.START': 1359, 'Q.END': 1382, 'T.START': 73307546, 'T.END': 73307569, 'CG': '23='}, {'Q.START': 1434, 'Q.END': 1451, 'T.START': 73307643, 'T.END': 73307660, 'CG': '17='}, {'Q.START': 1451, 'Q.END': 1531, 'T.START': 73307661, 'T.END': 73307741, 'CG': '80='}, {'Q.START': 1532, 'Q.END': 1543, 'T.START': 73307744, 'T.END': 73307755, 'CG': '11='}, {'Q.START': 1544, 'Q.END': 1572, 'T.START': 73307758, 'T.END': 73307786, 'CG': '28='}, {'Q.START': 1572, 'Q.END': 1573, 'T.START': 73307787, 'T.END': 73307788, 'CG': '1='}, {'Q.START': 1573, 'Q.END': 1587, 'T.START': 73307789, 'T.END': 73307803, 'CG': '14='}, {'Q.START': 1588, 'Q.END': 1616, 'T.START': 73307806, 'T.END': 73307834, 'CG': '28='}, {'Q.START': 1616, 'Q.END': 1617, 'T.START': 73307835, 'T.END': 73307836, 'CG': '1='}, {'Q.START': 1617, 'Q.END': 1646, 'T.START': 73307837, 'T.END': 73307866, 'CG': '29='}, {'Q.START': 1646, 'Q.END': 1661, 'T.START': 73307867, 'T.END': 73307882, 'CG': '15='}, {'Q.START': 1661, 'Q.END': 1673, 'T.START': 73307883, 'T.END': 73307895, 'CG': '12='}, {'Q.START': 1673, 'Q.END': 1674, 'T.START': 73307896, 'T.END': 73307897, 'CG': '1='}, {'Q.START': 1674, 'Q.END': 1726, 'T.START': 73307898, 'T.END': 73307950, 'CG': '52='}, {'Q.START': 1727, 'Q.END': 1762, 'T.START': 73307953, 'T.END': 73307988, 'CG': '35='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1767, 'Q.END': 1824, 'T.START': 73307995, 'T.END': 73308052, 'CG': '57='}, {'Q.START': 1824, 'Q.END': 1825, 'T.START': 73308053, 'T.END': 73308054, 'CG': '1='}, {'Q.START': 1825, 'Q.END': 1975, 'T.START': 73308055, 'T.END': 73308205, 'CG': '150='}, {'Q.START': 1976, 'Q.END': 2015, 'T.START': 73308208, 'T.END': 73308247, 'CG': '39='}, {'Q.START': 2016, 'Q.END': 2047, 'T.START': 73308250, 'T.END': 73308281, 'CG': '31='}, {'Q.START': 2047, 'Q.END': 2055, 'T.START': 73308286, 'T.END': 73308294, 'CG': '8='}, {'Q.START': 2056, 'Q.END': 2120, 'T.START': 73308297, 'T.END': 73308361, 'CG': '64='}, {'Q.START': 2120, 'Q.END': 2121, 'T.START': 73308362, 'T.END': 73308363, 'CG': '1='}, {'Q.START': 2121, 'Q.END': 2157, 'T.START': 73308364, 'T.END': 73308400, 'CG': '36='}, {'Q.START': 2158, 'Q.END': 2170, 'T.START': 73308403, 'T.END': 73308415, 'CG': '12='}, {'Q.START': 2170, 'Q.END': 2171, 'T.START': 73308416, 'T.END': 73308417, 'CG': '1='}, {'Q.START': 2171, 'Q.END': 2205, 'T.START': 73308418, 'T.END': 73308452, 'CG': '34='}, {'Q.START': 2206, 'Q.END': 2344, 'T.START': 73308455, 'T.END': 73308593, 'CG': '138='}, {'Q.START': 2345, 'Q.END': 2364, 'T.START': 73308596, 'T.END': 73308615, 'CG': '19='}, {'Q.START': 2364, 'Q.END': 2383, 'T.START': 73308616, 'T.END': 73308635, 'CG': '19='}, {'Q.START': 2383, 'Q.END': 2408, 'T.START': 73308636, 'T.END': 73308661, 'CG': '25='}, {'Q.START': 2408, 'Q.END': 2409, 'T.START': 73308662, 'T.END': 73308663, 'CG': '1='}, {'Q.START': 2409, 'Q.END': 2441, 'T.START': 73308664, 'T.END': 73308696, 'CG': '32='}, {'Q.START': 2441, 'Q.END': 2442, 'T.START': 73308697, 'T.END': 73308698, 'CG': '1='}, {'Q.START': 2442, 'Q.END': 2580, 'T.START': 73308699, 'T.END': 73308837, 'CG': '138='}, {'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}, {'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}, {'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}, {'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}, {'Q.START': 2584, 'Q.END': 2764, 'T.START': 73308842, 'T.END': 73309022, 'CG': '180='}, {'Q.START': 2765, 'Q.END': 2797, 'T.START': 73309025, 'T.END': 73309057, 'CG': '32='}, {'Q.START': 2798, 'Q.END': 2878, 'T.START': 73309060, 'T.END': 73309140, 'CG': '80='}, {'Q.START': 2878, 'Q.END': 2879, 'T.START': 73309141, 'T.END': 73309142, 'CG': '1='}, {'Q.START': 2879, 'Q.END': 2951, 'T.START': 73309143, 'T.END': 73309215, 'CG': '72='}, {'Q.START': 2951, 'Q.END': 2952, 'T.START': 73309216, 'T.END': 73309217, 'CG': '1='}, {'Q.START': 2952, 'Q.END': 3002, 'T.START': 73309218, 'T.END': 73309268, 'CG': '50='}, {'Q.START': 3002, 'Q.END': 3077, 'T.START': 73309271, 'T.END': 73309346, 'CG': '75='}, {'Q.START': 3077, 'Q.END': 3078, 'T.START': 73309347, 'T.END': 73309348, 'CG': '1='}, {'Q.START': 3078, 'Q.END': 3093, 'T.START': 73309349, 'T.END': 73309364, 'CG': '15='}, {'Q.START': 3094, 'Q.END': 3097, 'T.START': 73309367, 'T.END': 73309370, 'CG': '3='}, {'Q.START': 3097, 'Q.END': 3140, 'T.START': 73309371, 'T.END': 73309414, 'CG': '43='}, {'Q.START': 3140, 'Q.END': 3210, 'T.START': 73309415, 'T.END': 73309485, 'CG': '70='}, {'Q.START': 3210, 'Q.END': 3211, 'T.START': 73309486, 'T.END': 73309487, 'CG': '1='}, {'Q.START': 3211, 'Q.END': 3229, 'T.START': 73309488, 'T.END': 73309506, 'CG': '18='}, {'Q.START': 3229, 'Q.END': 3230, 'T.START': 73309507, 'T.END': 73309508, 'CG': '1='}, {'Q.START': 3230, 'Q.END': 3276, 'T.START': 73309509, 'T.END': 73309555, 'CG': '46='}, {'Q.START': 3277, 'Q.END': 3315, 'T.START': 73309558, 'T.END': 73309596, 'CG': '38='}, {'Q.START': 3316, 'Q.END': 3322, 'T.START': 73309599, 'T.END': 73309605, 'CG': '6='}, {'Q.START': 3323, 'Q.END': 3348, 'T.START': 73309608, 'T.END': 73309633, 'CG': '25='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3353, 'Q.END': 3354, 'T.START': 73309638, 'T.END': 73309639, 'CG': '1='}, {'Q.START': 3354, 'Q.END': 3356, 'T.START': 73309640, 'T.END': 73309642, 'CG': '2='}, {'Q.START': 3357, 'Q.END': 3489, 'T.START': 73309645, 'T.END': 73309777, 'CG': '132='}, {'Q.START': 3490, 'Q.END': 3642, 'T.START': 73309780, 'T.END': 73309932, 'CG': '152='}, {'Q.START': 3644, 'Q.END': 3685, 'T.START': 73309933, 'T.END': 73309974, 'CG': '41='}, {'Q.START': 3687, 'Q.END': 3693, 'T.START': 73309977, 'T.END': 73309983, 'CG': '6='}, {'Q.START': 3694, 'Q.END': 3708, 'T.START': 73309986, 'T.END': 73310000, 'CG': '14='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}, {'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}, {'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}]\n"
+ ]
+ }
+ ],
+ "source": [
+ "print(ALNS[(\"D134#1#chr03\", \"ALN_1\")])"
+ ]
+ }
+ ],
+ "metadata": {
+ "kernelspec": {
+ "display_name": "Python 3 (ipykernel)",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3",
+ "version": "3.10.14"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/gaf2aln.py b/gaf2aln.py
index 7657046..25fdebf 100644
--- a/gaf2aln.py
+++ b/gaf2aln.py
@@ -412,40 +412,30 @@ for aln_name in aln_dict.keys():
tmp_aln["CG"] += f"{nodes_length[node_id]}I"
elif len(_) and _[-1]["Q.END"] == q_start: # Following on the query, not on the target (i.e. Deletion)
tmp_aln["Q.END"] = q_end
- tmps_aln["CG"] += f"{nodes_length[node_id]}D"
+ tmp_aln["CG"] += f"{nodes_length[node_id]}D"
else : # Else, completely different
+ try :
+ _.append(tmp_aln)
+ except : pass
tmp_aln = {
"Q.START": q_start,
"Q.END": q_end,
"T.START": t_start,
"T.END": t_end,
"CG": _CG,
- }
- print("\t", tmp_aln)
-
+ }
+
else :
print("\tNot in path")
# Node is not in the path
- _.append(tmp_aln)
- ALNS[(path_name, aln_name)] = _
-
-## Debug
-for elem in ALNS[("TO1000#1#chr03", "ALN_1")]:
- print(elem)
-
+ ALNS[(path_name, aln_name)] = _
-
-
-
-
-
-
-
-
-
-
-
-
-
+## Debug
+for elem in ALNS.keys():
+ print(elem)
+
+for key, elem in ALNS.items():
+ print(key)
+ print(elem)
\ No newline at end of file
--
GitLab
From ba8d0c2bf543e8e82785974a8c4da89487304e75 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 13:36:27 +0200
Subject: [PATCH 16/30] Update Anchors2Path.py
---
Anchors2Path.py | 18 +++++++++++++-----
1 file changed, 13 insertions(+), 5 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index 3111c6f..e9582f8 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -121,12 +121,14 @@ node_path_count = {}
## {<NODE_ID>: <Number of path traversing this node>}
# Computing number of path traversing each nodes
for path_id, node_list in path_nodes.items():
- for node_id in node_list:
+ for node_id, counts in np.unique(node_list, return_counts=True):
- try :
- node_path_count[node_id] += 1
- except :
- node_path_count[node_id] = 1
+ # Filtering anchors based on appearance (we keep unique anchors)
+ if counts == 1 :
+ try :
+ node_path_count[node_id] += 1
+ except :
+ node_path_count[node_id] = 1
# Searching anchors
n_path = len(list(path_nodes.keys()))
@@ -137,17 +139,23 @@ for node_id, count in node_path_count.items():
# Computing path position for each node of the path of interest
current_pos = 0
+ordered_anchors = []
for node_id in path_nodes[args.pathname]:
_end = current_pos + nodes_length[node_id]
# Trying to add anchors path position if it is an anchor
try :
Anchors[int(node_id)].append( (current_pos, _end) )
+
+ # Keeping track of order of appearance of anchors
+ ordered_anchors.append(int(node_id))
except:
pass
current_pos = _end
+# Filtering anchors based on
+
# Transforming data into a table
ID, START, END = [], [], []
for node_id, positions in Anchors.items():
--
GitLab
From 022035307bd73924f77b2e7b0995bafe03210e30 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 13:42:34 +0200
Subject: [PATCH 17/30] Update Anchors2Path.py
---
Anchors2Path.py | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index e9582f8..8a8261d 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -121,10 +121,11 @@ node_path_count = {}
## {<NODE_ID>: <Number of path traversing this node>}
# Computing number of path traversing each nodes
for path_id, node_list in path_nodes.items():
- for node_id, counts in np.unique(node_list, return_counts=True):
-
+ nodes_counts = np.unique(node_list, return_counts=True)
+ for i in range(len(nodes_counts)):
+ node_id, count = nodes_counts[0][i], nodes_counts[1][i]
# Filtering anchors based on appearance (we keep unique anchors)
- if counts == 1 :
+ if count == 1 :
try :
node_path_count[node_id] += 1
except :
--
GitLab
From 8a62837dad2792b87c61deb44cc1927f0dca03a9 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 14:48:26 +0200
Subject: [PATCH 18/30] Update Anchors2Path.py
---
Anchors2Path.py | 1 +
1 file changed, 1 insertion(+)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index 8a8261d..a3df05c 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -124,6 +124,7 @@ for path_id, node_list in path_nodes.items():
nodes_counts = np.unique(node_list, return_counts=True)
for i in range(len(nodes_counts)):
node_id, count = nodes_counts[0][i], nodes_counts[1][i]
+ print(node_id, count)
# Filtering anchors based on appearance (we keep unique anchors)
if count == 1 :
try :
--
GitLab
From 1a0dd8fb93808ec443394ae2eaf4eaf50955e5a4 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 15:02:06 +0200
Subject: [PATCH 19/30] Update Anchors2Path.py
---
Anchors2Path.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index a3df05c..98565e4 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -121,11 +121,11 @@ node_path_count = {}
## {<NODE_ID>: <Number of path traversing this node>}
# Computing number of path traversing each nodes
for path_id, node_list in path_nodes.items():
+ print(path_id)
nodes_counts = np.unique(node_list, return_counts=True)
for i in range(len(nodes_counts)):
node_id, count = nodes_counts[0][i], nodes_counts[1][i]
- print(node_id, count)
- # Filtering anchors based on appearance (we keep unique anchors)
+ print("\t", node_id, count)
if count == 1 :
try :
node_path_count[node_id] += 1
--
GitLab
From 291de812648d640e807678e771f2198153cc64f0 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 15:05:03 +0200
Subject: [PATCH 20/30] Update Anchors2Path.py
---
Anchors2Path.py | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index 98565e4..40c119f 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -17,7 +17,7 @@ from functools import reduce
import concurrent.futures
import gzip
-version = "0.1"
+version = "0.1.1"
## Argument parser
arg_parser = argparse.ArgumentParser(description='Anchors2Path')
@@ -72,6 +72,8 @@ if args.version:
print(version)
os._exit(0)
+print("Version:\t", version)
+
# Timing the script
start_time = time.time()
@@ -121,9 +123,10 @@ node_path_count = {}
## {<NODE_ID>: <Number of path traversing this node>}
# Computing number of path traversing each nodes
for path_id, node_list in path_nodes.items():
- print(path_id)
+
nodes_counts = np.unique(node_list, return_counts=True)
- for i in range(len(nodes_counts)):
+ print(path_id)
+ for i in range(len(nodes_counts[0])):
node_id, count = nodes_counts[0][i], nodes_counts[1][i]
print("\t", node_id, count)
if count == 1 :
--
GitLab
From 02a3e63c343d0dc0123c49ba5fe5b46dcbbc0aa8 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 3 Jun 2024 15:07:48 +0200
Subject: [PATCH 21/30] Update Anchors2Path.py
---
Anchors2Path.py | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/Anchors2Path.py b/Anchors2Path.py
index 40c119f..eb325ec 100644
--- a/Anchors2Path.py
+++ b/Anchors2Path.py
@@ -17,7 +17,7 @@ from functools import reduce
import concurrent.futures
import gzip
-version = "0.1.1"
+version = "0.1.2"
## Argument parser
arg_parser = argparse.ArgumentParser(description='Anchors2Path')
@@ -125,10 +125,10 @@ node_path_count = {}
for path_id, node_list in path_nodes.items():
nodes_counts = np.unique(node_list, return_counts=True)
- print(path_id)
+ print(f"[Anchors2Path] Counting nodes in {path_id}")
for i in range(len(nodes_counts[0])):
node_id, count = nodes_counts[0][i], nodes_counts[1][i]
- print("\t", node_id, count)
+
if count == 1 :
try :
node_path_count[node_id] += 1
--
GitLab
From c4ff0c313ab30d67b4d66086ac6ccad16f99b2ed Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Thu, 27 Jun 2024 17:11:32 +0200
Subject: [PATCH 22/30] Adding UniP
---
GFAvc.py | 6 ++----
PanGeTools.def | 4 ++++
2 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/GFAvc.py b/GFAvc.py
index 7ac2173..5cf601f 100644
--- a/GFAvc.py
+++ b/GFAvc.py
@@ -44,6 +44,7 @@ if args.version:
with open(args.gfa, 'r') as file:
gfa = file.readlines()
+ # for line in file :
## Changing version number in header
assert gfa[0].split('\t')[1] == "VN:Z:1.1"
@@ -59,10 +60,7 @@ for lineID in sorted(range(len(gfa)), reverse = True):
if gfa[lineID][0] == "W" :
- curLine = gfa.pop(lineID).split('\t')
-
- # Removing '\n' at the end of the line
- curLine[-1] = curLine[-1][:-1]
+ curLine = gfa.pop(lineID).strip().split('\t')
# Transforming '>..>..<..>..' to ['>..', '>..', '<..', '>..']
curWalk = re.findall(r'>\w+|<\w+', curLine[-1])
diff --git a/PanGeTools.def b/PanGeTools.def
index 41e76ac..0e0a29c 100644
--- a/PanGeTools.def
+++ b/PanGeTools.def
@@ -209,6 +209,10 @@ Stage: build
mv bin/GraphAligner /apps/bin/
cd /apps && rm -rf GraphAligner
+ # Installing UniP
+ cd /apps
+ git clone https://forgemia.inra.fr/alexis.mergez/unip.git UniP
+
# Cleaning step
## Removing tarballs
rm /apps/*.tar*
--
GitLab
From 734e3396599ba5ab710bfc74ee5232899df1aac8 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Tue, 9 Jul 2024 20:07:20 +0200
Subject: [PATCH 23/30] Linked Unip in container
---
PanGeTools.def | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/PanGeTools.def b/PanGeTools.def
index 0e0a29c..1f850ac 100644
--- a/PanGeTools.def
+++ b/PanGeTools.def
@@ -212,11 +212,15 @@ Stage: build
# Installing UniP
cd /apps
git clone https://forgemia.inra.fr/alexis.mergez/unip.git UniP
+ ln -s /apps/UniP/UniP.py /apps/bin/UniP.py
# Cleaning step
## Removing tarballs
rm /apps/*.tar*
+%apprun unip
+ exec UniP.py "$@"
+
%apprun gfaffix
exec gfaffix "$@"
--
GitLab
From 079dbc0ccf7aa9ec634d3dd288e055d2ca064c78 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Tue, 9 Jul 2024 20:07:49 +0200
Subject: [PATCH 24/30] Created Pancat unofficial container
---
.gitlab-ci.yml | 30 +++++++++++++++++++++++++++++-
Pancat.def | 36 ++++++++++++++++++++++++++++++++++++
2 files changed, 65 insertions(+), 1 deletion(-)
create mode 100644 Pancat.def
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 9068aa4..834c834 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -39,9 +39,19 @@ test_for_changes:
else
RUN_panache="0";
fi
+
+ # Check block for Pancat.def
+ - curMd5=$(git show ${curTag}:Pancat.def | md5sum | cut -d' ' -f1)
+ - prevMd5=$(git show ${prevTag}:Pancat.def | md5sum | cut -d' ' -f1)
+ - if [ $curMd5 != $prevMd5 ]; then
+ RUN_Pancat="1";
+ else
+ RUN_Pancat="0";
+ fi
- echo "RUN_PanGeTools=$RUN_PanGeTools" >> build.env
- echo "RUN_panache=$RUN_panache" >> build.env
+ - echo "RUN_Pancat=$RUN_Pancat" >> build.env
artifacts:
reports:
@@ -87,4 +97,22 @@ build:panache:
apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" panache.sif oras://"$CI_REGISTRY_IMAGE"/panache:latest;
fi
-
+build:Pancat:
+ stage: build
+ tags:
+ - stable # Using Stable runners as test runners don't work properly
+ needs:
+ - job: "test_for_changes"
+ artifacts: true
+ image:
+ name: kaczmarj/apptainer:latest
+ entrypoint: [""]
+ rules:
+ - if: $CI_COMMIT_TAG # Run this job when a tag is created
+ script:
+ - if [ $RUN_Pancat == "1" ]; then
+ sed -i "s/VERSION_NUMBER/${CI_COMMIT_TAG#v}/" Pancat.def;
+ apptainer build Pancat.sif Pancat.def;
+ apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" Pancat.sif oras://"$CI_REGISTRY_IMAGE"/pancat:"$CI_COMMIT_TAG";
+ apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" Pancat.sif oras://"$CI_REGISTRY_IMAGE"/pancat:latest;
+ fi
diff --git a/Pancat.def b/Pancat.def
new file mode 100644
index 0000000..67aac3d
--- /dev/null
+++ b/Pancat.def
@@ -0,0 +1,36 @@
+Bootstrap: docker
+From: ghcr.io/mamba-org/micromamba:latest
+Stage: build
+
+%environment
+ export MAMBA_DOCKERFILE_ACTIVATE=1
+ export PATH="$PATH:/apps/Pancat"
+
+%post
+
+ export MAMBA_DOCKERFILE_ACTIVATE=1
+ apt-get update && apt-get upgrade -y
+ apt-get install -y git
+
+ # Creating base environment with micromamba
+ mkdir /apps
+ micromamba create -y -p /apps/base -c conda-forge -c bioconda \
+ python=3.10.*
+
+ git clone https://github.com/Tharos-ux/pancat.git /apps/Pancat
+ cd /apps/Pancat
+ micromamba run -p /apps/base pip install -r requirements.txt --upgrade
+ micromamba run -p /apps/base python -m pip install . --quiet
+
+%runscript
+ exec micromamba run -p /apps/base "$@"
+
+%labels
+ Author alexis.mergez@inrae.fr
+ Image.version VERSION_NUMBER
+ pancat.home https://github.com/Tharos-ux/pancat.git
+ about.home https://forgemia.inra.fr/alexis.mergez/pan1capps
+
+%help
+ Unofficial Apptainer container for Pancat (https://github.com/Tharos-ux/pancat.git).
+ Image version : VERSION_NUMBER
--
GitLab
From db6fa3107a42e3fb6204ca7aacb4541878876a22 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Tue, 9 Jul 2024 20:10:51 +0200
Subject: [PATCH 25/30] Tricking CICD
---
Pancat.def | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/Pancat.def b/Pancat.def
index 67aac3d..f87205f 100644
--- a/Pancat.def
+++ b/Pancat.def
@@ -7,7 +7,6 @@ Stage: build
export PATH="$PATH:/apps/Pancat"
%post
-
export MAMBA_DOCKERFILE_ACTIVATE=1
apt-get update && apt-get upgrade -y
apt-get install -y git
@@ -33,4 +32,4 @@ Stage: build
%help
Unofficial Apptainer container for Pancat (https://github.com/Tharos-ux/pancat.git).
- Image version : VERSION_NUMBER
+ Image version: VERSION_NUMBER
--
GitLab
From b20d7852034afeeb4cc03f76ba4f45361623fa05 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 26 Jul 2024 12:00:11 +0200
Subject: [PATCH 26/30] Update Pancat.def
---
Pancat.def | 1 +
1 file changed, 1 insertion(+)
diff --git a/Pancat.def b/Pancat.def
index f87205f..56ba7e0 100644
--- a/Pancat.def
+++ b/Pancat.def
@@ -33,3 +33,4 @@ Stage: build
%help
Unofficial Apptainer container for Pancat (https://github.com/Tharos-ux/pancat.git).
Image version: VERSION_NUMBER
+
--
GitLab
From d8cfcfcefe79d85b48b6038eb56ca89b2295ae7a Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Thu, 1 Aug 2024 17:59:44 +0200
Subject: [PATCH 27/30] Update PanGeTools.def
- Added Seqwish v0.7.9
- Updated VG to v1.58.0
---
PanGeTools.def | 19 ++++++++++++++++---
1 file changed, 16 insertions(+), 3 deletions(-)
diff --git a/PanGeTools.def b/PanGeTools.def
index 1f850ac..79465d8 100644
--- a/PanGeTools.def
+++ b/PanGeTools.def
@@ -96,7 +96,7 @@ Stage: build
# Installing vg
cd /apps/bin
- wget --no-check-certificate -c https://github.com/vgteam/vg/releases/download/v1.56.0/vg
+ wget --no-check-certificate -c https://github.com/vgteam/vg/releases/download/v1.58.0/vg
chmod +x vg
# Installing panacus
@@ -126,6 +126,14 @@ Stage: build
cmake -H. -Bbuild && cmake --build build -- -j $(nproc)
mv /apps/wfmash-v0.10.5/build/bin/wfmash /apps/bin/wfmash
+ # Installing seqwish
+ cd /apps
+ wget --no-check-certificate https://github.com/ekg/seqwish/releases/download/v0.7.9/seqwish-v0.7.9.tar.gz
+ tar -zxvf seqwish-v0.7.9.tar.gz
+ cd seqwish-v0.7.9
+ cmake -H. -Bbuild && cmake --build build -- -j $(nproc)
+ mv /apps/seqwish-v0.7.9/bin/seqwish /apps/bin/seqwish
+
# Installing bgzip
cd /apps
wget --no-check-certificate https://github.com/samtools/htslib/releases/download/1.19.1/htslib-1.19.1.tar.bz2
@@ -257,6 +265,9 @@ Stage: build
%apprun wfmash
exec wfmash "$@"
+%apprun seqwish
+ exec seqwish "$@"
+
%apprun minimap2
exec minimap2 "$@"
@@ -320,13 +331,14 @@ Stage: build
GFAffix.Version 0.1.5
smoothxg.Version 0.7.2
Odgi.Version 0.8.6
- vg.Version 1.56.0
+ vg.Version 1.58.0
panacus.Version 0.2.3
gfatools.Version 0.5
GFAvc.Version 0.3
GFAstats.Version 0.3.2
Samtools.Version 1.19
wfmash.Version 0.10.5
+ seqwish.Version 0.7.9
htslib.Version 1.19.1
minimap2.Version 2.26
vcfbub.Version 0.1.0
@@ -344,7 +356,7 @@ Stage: build
- GFAffix v0.1.5
- smoothxg v0.7.2
- odgi v0.8.6
- - vg v1.56.0
+ - vg v1.58.0
- panacus v0.2.3
- gfatools v0.5
- GFAvc v0.3
@@ -352,6 +364,7 @@ Stage: build
- samtools v1.19
- bgzip v1.19.1
- wfmash v0.10.5
+ - seqwish v0.7.9
- minimap2 v2.26
- vcfbub v0.1.0
- vcflib v1.0.9
--
GitLab
From faf2580e937979f6e2baaaf1c4ba790bcb64196b Mon Sep 17 00:00:00 2001
From: Alexis Mergez <alexis.mergez@inrae.fr>
Date: Fri, 2 Aug 2024 16:55:16 +0200
Subject: [PATCH 28/30] Bumped Wfmash version
---
.gitignore | 3 +-
.gitlab-ci.yml | 119 ++++++++++++++-----------------------------------
PanGeTools.def | 12 ++---
3 files changed, 42 insertions(+), 92 deletions(-)
diff --git a/.gitignore b/.gitignore
index 1fcf15d..75c3153 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1,2 @@
-*.sif
\ No newline at end of file
+*.sif
+PGT-beta.def
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 834c834..40251c1 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -1,118 +1,67 @@
-release_job:
- stage: .pre
- image: registry.gitlab.com/gitlab-org/release-cli:latest
- rules:
- - if: $CI_COMMIT_TAG # Run this job when a tag is created
- script:
- - echo "running release_job"
- release: # See https://docs.gitlab.com/ee/ci/yaml/#release for available properties
- tag_name: '$CI_COMMIT_TAG'
- description: '$CI_COMMIT_TAG'
-
-test_for_changes:
- stage: .pre
+Build_PanGeTools:
+ stage: build
tags:
- stable # Using Stable runners as test runners don't work properly
- image: ubuntu:latest
+ image:
+ name: kaczmarj/apptainer:latest
+ entrypoint: [""]
rules:
- - if: $CI_COMMIT_TAG
+ - if: $CI_COMMIT_TAG # Run this job when a tag is created
+ when: manual
script:
- - apt update -y && apt install -y git
- # Getting current and previous tags
- - curTag=$(git describe --abbrev=0 --tags)
- - prevTag=$(git describe --abbrev=0 --tags HEAD^)
-
- # Check block for PanGeTools.def
- - curMd5=$(git show ${curTag}:PanGeTools.def | md5sum | cut -d' ' -f1)
- - prevMd5=$(git show ${prevTag}:PanGeTools.def | md5sum | cut -d' ' -f1)
- - if [ $curMd5 != $prevMd5 ]; then
- RUN_PanGeTools="1";
- else
- RUN_PanGeTools="0";
- fi
-
- # Check block for panache.def
- - curMd5=$(git show ${curTag}:Panache.def | md5sum | cut -d' ' -f1)
- - prevMd5=$(git show ${prevTag}:Panache.def | md5sum | cut -d' ' -f1)
- - if [ $curMd5 != $prevMd5 ]; then
- RUN_panache="1";
- else
- RUN_panache="0";
- fi
-
- # Check block for Pancat.def
- - curMd5=$(git show ${curTag}:Pancat.def | md5sum | cut -d' ' -f1)
- - prevMd5=$(git show ${prevTag}:Pancat.def | md5sum | cut -d' ' -f1)
- - if [ $curMd5 != $prevMd5 ]; then
- RUN_Pancat="1";
- else
- RUN_Pancat="0";
- fi
-
- - echo "RUN_PanGeTools=$RUN_PanGeTools" >> build.env
- - echo "RUN_panache=$RUN_panache" >> build.env
- - echo "RUN_Pancat=$RUN_Pancat" >> build.env
-
+ - sed -i "s/VERSION_NUMBER/${CI_COMMIT_TAG#v}/" PanGeTools.def
+ - apptainer build PanGeTools.sif PanGeTools.def
+ - apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" PanGeTools.sif oras://"$CI_REGISTRY_IMAGE"/pangetools:"$CI_COMMIT_TAG"
artifacts:
- reports:
- dotenv: build.env
+ paths:
+ - PanGeTools.sif
+ expire_in: 1 week
-build:PanGeTools:
- stage: build
+Latest_PanGeTools:
+ stage: deploy
tags:
- stable # Using Stable runners as test runners don't work properly
needs:
- - job: "test_for_changes"
- artifacts: true
+ - job: "Build_PanGeTools"
image:
name: kaczmarj/apptainer:latest
entrypoint: [""]
rules:
- - if: $CI_COMMIT_TAG # Run this job when a tag is created
+ - if: $CI_COMMIT_TAG # Run this job when a tag is created
+ when: manual
script:
- - if [ $RUN_PanGeTools == "1" ]; then
- sed -i "s/VERSION_NUMBER/${CI_COMMIT_TAG#v}/" PanGeTools.def;
- apptainer build PanGeTools.sif PanGeTools.def;
- apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" PanGeTools.sif oras://"$CI_REGISTRY_IMAGE"/pangetools:"$CI_COMMIT_TAG";
- apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" PanGeTools.sif oras://"$CI_REGISTRY_IMAGE"/pangetools:latest;
- fi
+ - apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" PanGeTools.sif oras://"$CI_REGISTRY_IMAGE"/pangetools:latest
-build:panache:
+Build_Pancat:
stage: build
tags:
- stable # Using Stable runners as test runners don't work properly
- needs:
- - job: "test_for_changes"
- artifacts: true
image:
name: kaczmarj/apptainer:latest
entrypoint: [""]
rules:
- if: $CI_COMMIT_TAG # Run this job when a tag is created
+ when: manual
script:
- - if [ $RUN_panache == "1" ]; then
- sed -i "s/VERSION_NUMBER/${CI_COMMIT_TAG#v}/" Panache.def;
- apptainer build Panache.sif Panache.def;
- apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" panache.sif oras://"$CI_REGISTRY_IMAGE"/panache:"$CI_COMMIT_TAG";
- apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" panache.sif oras://"$CI_REGISTRY_IMAGE"/panache:latest;
- fi
+ - sed -i "s/VERSION_NUMBER/${CI_COMMIT_TAG#v}/" Pancat.def
+ - apptainer build Pancat.sif Pancat.def
+ - apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" Pancat.sif oras://"$CI_REGISTRY_IMAGE"/pancat:"$CI_COMMIT_TAG"
+ artifacts:
+ paths:
+ - Pancat.sif
+ expire_in: 1 week
-build:Pancat:
- stage: build
+Latest_Pancat:
+ stage: deploy
tags:
- stable # Using Stable runners as test runners don't work properly
needs:
- - job: "test_for_changes"
- artifacts: true
+ - job: "Build_Pancat"
image:
name: kaczmarj/apptainer:latest
entrypoint: [""]
rules:
- - if: $CI_COMMIT_TAG # Run this job when a tag is created
+ - if: $CI_COMMIT_TAG # Run this job when a tag is created
+ when: manual
script:
- - if [ $RUN_Pancat == "1" ]; then
- sed -i "s/VERSION_NUMBER/${CI_COMMIT_TAG#v}/" Pancat.def;
- apptainer build Pancat.sif Pancat.def;
- apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" Pancat.sif oras://"$CI_REGISTRY_IMAGE"/pancat:"$CI_COMMIT_TAG";
- apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" Pancat.sif oras://"$CI_REGISTRY_IMAGE"/pancat:latest;
- fi
+ - apptainer push --docker-username "${CI_REGISTRY_USER}" --docker-password "${CI_REGISTRY_PASSWORD}" Pancat.sif oras://"$CI_REGISTRY_IMAGE"/pancat:latest
\ No newline at end of file
diff --git a/PanGeTools.def b/PanGeTools.def
index 79465d8..ecc05f1 100644
--- a/PanGeTools.def
+++ b/PanGeTools.def
@@ -120,11 +120,11 @@ Stage: build
# Installing wfmash
cd /apps
- wget --no-check-certificate https://github.com/waveygang/wfmash/releases/download/v0.10.5/wfmash-v0.10.5.tar.gz
- tar -zxvf wfmash-v0.10.5.tar.gz
- cd wfmash-v0.10.5
+ wget --no-check-certificate https://github.com/waveygang/wfmash/releases/download/v0.17.0/wfmash-v0.17.0.tar.gz
+ tar -zxvf wfmash-v0.17.0.tar.gz
+ cd wfmash-v0.17.0
cmake -H. -Bbuild && cmake --build build -- -j $(nproc)
- mv /apps/wfmash-v0.10.5/build/bin/wfmash /apps/bin/wfmash
+ mv /apps/wfmash-v0.17.0/build/bin/wfmash /apps/bin/wfmash
# Installing seqwish
cd /apps
@@ -337,7 +337,7 @@ Stage: build
GFAvc.Version 0.3
GFAstats.Version 0.3.2
Samtools.Version 1.19
- wfmash.Version 0.10.5
+ wfmash.Version 0.17.0
seqwish.Version 0.7.9
htslib.Version 1.19.1
minimap2.Version 2.26
@@ -363,7 +363,7 @@ Stage: build
- GFAstats v0.3.2
- samtools v1.19
- bgzip v1.19.1
- - wfmash v0.10.5
+ - wfmash v0.17.0
- seqwish v0.7.9
- minimap2 v2.26
- vcfbub v0.1.0
--
GitLab
From c77499c2e10b85b05cfebf6fc7b0556bbb7b70f1 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 5 Aug 2024 17:54:08 +0200
Subject: [PATCH 29/30] GFAvc v0.4
Changelog :
- Added support for gzipped GFA
- Stripping escape character when parsing
- Added GFA1.0 to GFA1.1 conversion direction (intended for PGGB GFA)
- Optional index can be passed to add START-END range to walks when converting from GFA1.0 to GFA1.0
Note on GFA1.0 to GF1.1 conversion :
As PGGB has no reference, every sample is set as reference (i.e. added to the space separated list from the GFA header under RS:Z:...). This simplify the use of VG tools such as vg surject which only surject onto reference walks. This also prevent paths/walks from having different names between vg and non-vg file format (no phase_block)
---
GFAvc.py | 187 ++++++++++++++++++++++++++++++++++++++++++++-----------
1 file changed, 150 insertions(+), 37 deletions(-)
diff --git a/GFAvc.py b/GFAvc.py
index 5cf601f..f130f10 100644
--- a/GFAvc.py
+++ b/GFAvc.py
@@ -2,33 +2,45 @@
# -*- coding: utf-8 -*-
"""
GFAvc: GFA version converter.
-Convert GFA from v1.1 to v1.0 (Convert walk to paths)
+Convert GFA from v1.1 to v1.0 (Convert walk to paths) and vice versa.
@author: alexis.mergez@inrae.fr
-@version: 0.3
+@version: 0.4
"""
import re
import argparse
import os
+import gzip
-version = "0.3"
+version = "0.4"
## Argument parser
arg_parser = argparse.ArgumentParser(description='GFAvc: GFA version converter')
arg_parser.add_argument(
- "--gfa",
+ "--gfa1",
"-g",
- dest = "gfa",
- required = True,
- help = "GFA 1.1 file."
- )
+ dest = "GFA1",
+ help = "GFA 1.1 file. (Gzip or not)"
+ )
+arg_parser.add_argument(
+ "--gfa",
+ "-G",
+ dest = "GFA",
+ help = "GFA 1.0 file. (Gzip or not)"
+ )
arg_parser.add_argument(
"--outName",
"-o",
dest = "outName",
required = True,
help = "Output file name."
- )
+ )
+arg_parser.add_argument(
+ "--index",
+ "-i",
+ dest = "index",
+ help = "TSV containing start stop for each path in order to convert to walk (optional)"
+ )
arg_parser.add_argument(
'--version',
'-v',
@@ -38,45 +50,146 @@ arg_parser.add_argument(
)
args = arg_parser.parse_args()
+#% Returning version
if args.version:
print(version)
- os._exit(0)
+ os._exit(0)
+
+#% Parsing index
+if args.index is not None:
+ with open(args.index, 'r') as handle:
+ file = [line.rstrip() for line in handle.readlines()]
+
+ index = {}
+ for line in file:
+ split = line.split("\t")
+ index[split[0]] = [split[1], split[2]]
+
+else : index = None
+
+#% Conversion functions
+def gfa11_to_gfa10(gfa1_file = args.GFA1):
+ #% Reading GFA
+ # If not gzipped :
+ if gfa1_file[-2:] != "gz" :
+ with open(gfa1_file, 'r') as file:
+ gfa = [line.rstrip() for line in file.readlines()]
+
+ # If gzipped :
+ else :
+ with gzip.open(gfa1_file, 'r') as file:
+ gfa = [line.decode().rstrip() for line in file.readlines()]
+
+ #% Changing version number in header
+ assert gfa[0].split('\t')[1] == "VN:Z:1.1"
+ _ = gfa[0].split('\t')
+ _[1] = "VN:Z:1.0"
+ gfa[0] = "\t".join(_)
+
+ #% Iterating in reverse to put paths at the end.
+ for lineID in sorted(range(len(gfa)), reverse = True):
+ if gfa[lineID][0] == "S" :
+ _ = '\t'.join(gfa[lineID].split('\t')[:3])
+ gfa[lineID] = f"{_}"
+
+ if gfa[lineID][0] == "W" :
+
+ curLine = gfa.pop(lineID).split('\t')
+
+ # Transforming '>..>..<..>..' to ['>..', '>..', '<..', '>..']
+ curWalk = re.findall(r'>\w+|<\w+', curLine[-1])
+
+ # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'
+ path = [f'{elem[1:]}{(elem[0] == ">")*"+"+(elem[0] == "<")*"-"}' for elem in curWalk]
+
+ newLine = ['P', f"{'#'.join(curLine[1:5])}-{curLine[5]}", ','.join(path), '*']
+
+ gfa.append('\t'.join(newLine))
+
+ #% Moving path lines to the end
+ if gfa[lineID][0] == "P" :
+
+ curLine = gfa.pop(lineID)
+ gfa.append(curLine)
+
+ return gfa
+
+def gfa10_to_gfa11(gfa_file = args.GFA, index = index):
+
+ #% Reading GFA
+ # If not gzipped :
+ if gfa_file[-2:] != "gz" :
+ with open(gfa_file, 'r') as file:
+ gfa = [line.rstrip() for line in file.readlines()]
+
+ # If gzipped :
+ else :
+ with gzip.open(gfa_file, 'r') as file:
+ gfa = [line.decode().rstrip() for line in file.readlines()]
+
+ sign = {"+": ">", "-": "<"}
+ samples = []
+
+ #% Iterating in reverse to put walks at the end.
+ for lineID in sorted(range(len(gfa)), reverse = True):
+
+ if gfa[lineID][0] == "P" :
+
+ curLine = gfa.pop(lineID).split('\t')
-with open(args.gfa, 'r') as file:
- gfa = file.readlines()
- # for line in file :
+ # Converting '..+,..+,..-,..+' to ['>..', '>..', '<..', '>..']
+ walk = [
+ f"{sign[elem[-1]]}{elem[:-1]}"
+ for elem in curLine[-2].split(',')
+ ]
-## Changing version number in header
-assert gfa[0].split('\t')[1] == "VN:Z:1.1"
-_ = gfa[0].split('\t')
-_[1] = "VN:Z:1.0"
-gfa[0] = "\t".join(_)
+ splittedID = curLine[1].split(":")
+ if len(splittedID) == 2:
+ # Range info is available
+ ID = splittedID[0].split("#")
+ RANGE = splittedID[1].split("-")
+ elif len(splittedID) == 1 and index is not None:
+ ID = splittedID[0].split("#")
+ RANGE = index[splittedID[0]]
+ else :
+ ID = splittedID[0].split("#")
+ RANGE = 2*["*"]
+
+ newLine = ['W'] + ID + RANGE + [f"{''.join(walk)}"]
+ samples.append(ID[0])
-## Iterating in reverse to put paths at the end.
-for lineID in sorted(range(len(gfa)), reverse = True):
- if gfa[lineID][0] == "S" :
- _ = '\t'.join(gfa[lineID].split('\t')[:3])
- gfa[lineID] = f"{_}\n"
+ gfa.append('\t'.join(newLine))
- if gfa[lineID][0] == "W" :
+ #% Moving walk lines to the end
+ if gfa[lineID][0] == "W" :
- curLine = gfa.pop(lineID).strip().split('\t')
-
- # Transforming '>..>..<..>..' to ['>..', '>..', '<..', '>..']
- curWalk = re.findall(r'>\w+|<\w+', curLine[-1])
+ curLine = gfa.pop(lineID)
+ gfa.append(curLine)
- # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'
- path = [f'{elem[1:]}{(elem[0] == ">")*"+"+(elem[0] == "<")*"-"}' for elem in curWalk]
+ samples = list(set(samples))
+ #% Changing version number in header
+ assert gfa[0].split('\t')[1] == "VN:Z:1.0"
+ _ = gfa[0].split('\t')
+ _[1] = "VN:Z:1.1"
+ _.append(f"RS:Z:{' '.join(samples)}")
+ gfa[0] = "\t".join(_)
- newLine = ['P', f"{'#'.join(curLine[1:5])}-{curLine[5]}", ','.join(path), '*\n']
+ return gfa
- gfa.append('\t'.join(newLine))
+#% Selecting the conversion direction
+if args.GFA is None and args.GFA1 is not None:
+ print("[GFAvc] Converting from GFA 1.1 to GFA 1.0 ...")
+ gfa = gfa11_to_gfa10()
-## Moving path lines to the end
- if gfa[lineID][0] == "P" :
+elif args.GFA is not None and args.GFA1 is None and args.refname is not None:
+ print("[GFAvc] Converting from GFA 1.0 to GFA 1.1 ...")
+ gfa = gfa10_to_gfa11()
- curLine = gfa.pop(lineID)
- gfa.append(curLine)
+else:
+ print("[GFAvc] Unable to convert !")
+ os._exit(1)
+#% Exporting
+gfa[-1] = f"{gfa[-1]}\n"
with open(args.outName, "w") as file:
- file.write("".join(gfa))
\ No newline at end of file
+ file.write("\n".join(gfa))
\ No newline at end of file
--
GitLab
From 76a571bbe0328bb858451b95e61d7d88c707a1d8 Mon Sep 17 00:00:00 2001
From: Alexis Mergez <amergez@miat-pret-5.toulouse.inrae.fr>
Date: Mon, 5 Aug 2024 17:54:21 +0200
Subject: [PATCH 30/30] Removing/moving old scripts
---
DotPlot_BED_corrector.py | 129 --
gaf2aln.ipynb | 2443 --------------------------------------
gaf2aln.py | 441 -------
3 files changed, 3013 deletions(-)
delete mode 100644 DotPlot_BED_corrector.py
delete mode 100644 gaf2aln.ipynb
delete mode 100644 gaf2aln.py
diff --git a/DotPlot_BED_corrector.py b/DotPlot_BED_corrector.py
deleted file mode 100644
index f4858af..0000000
--- a/DotPlot_BED_corrector.py
+++ /dev/null
@@ -1,129 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-DotPlot Bed like file corrector.
-Correct splitted paths from odgi untangle bed like file used for creating dotplots.
-See Odgi documentation for the dot plot tutorial
-
-@author: alexis.mergez@inrae.fr
-@version: 0.1
-"""
-import re
-import argparse
-import os
-import numpy as np
-import time
-import pandas as pd
-from functools import reduce
-import concurrent.futures
-import gzip
-
-version = "0.1"
-
-## Argument parser
-arg_parser = argparse.ArgumentParser(description='GFAstats: GFA statistics')
-arg_parser.add_argument(
- "--input",
- "-i",
- dest = "input",
- required = True,
- help = "Bed like file"
- )
-arg_parser.add_argument(
- "--output",
- "-o",
- dest = "output",
- required = True,
- help = "Output name"
- )
-arg_parser.add_argument(
- '--version',
- '-v',
- action="store_true",
- dest = "version",
- help = "Show version"
-)
-arg_parser.add_argument(
- '--progress',
- '-P',
- action="store_true",
- dest = "progress",
- help = "Show progress to stdout"
-)
-args = arg_parser.parse_args()
-
-# Printing version and exiting if required
-if args.version:
- print(version)
- os._exit(0)
-
-# importing bed file with pandas
-if args.progress : print(f"[Bed_corrector::Parsing] Reading {args.input} ...")
-bed = pd.read_csv(
- args.input,
- sep = '\t'
-)
-
-# Getting the queries name and checking if we get multiples for one path
-if args.progress : print(f"[Bed_corrector::Identify] Searching for splitted paths ...")
-## Getting unique query names
-queries = bed["query.name"].unique()
-
-## Extracting path name and ranges from unique queries names
-paths = [query.split(":")[0] for query in queries]
-ranges = np.array([query.split(":")[1].split("-") for query in queries])
-
-## Creating temporary dataframe to store previous info
-temp_df = pd.DataFrame({
- "queries" : queries,
- "path" : paths,
- "start" : ranges[:, 0],
- "end" : ranges[:, 1]
-})
-
-## Getting the path that are splitted (i.e. more than one unique occurence)
-splitted_paths = []
-_ = np.unique(paths, return_counts = True)
-for path, count in zip(_[0], _[1]):
- if count > 1 : # More than a repeat
- splitted_paths.append(path)
- if args.progress : print(f"[Bed_corrector::Identify] {path} is splitted")
-
-# For each splitted path identified, we search the minimum start and the maximum end
-if args.progress : print(f"[Bed_corrector::Identify] Searching for min start and max end of splitted paths ...")
-min_start = {}
-max_end = {}
-for path_name in splitted_paths:
- min_start[path_name] = int(temp_df[temp_df.path == path_name].start.min())
- max_end[path_name] = int(temp_df[temp_df.path == path_name].end.max())
-
-# Traversing bed dataframe and offsetting paths based on the min_start of the main path
-if args.progress : print(f"[Bed_corrector::Patching] Correcting splitted paths ...")
-## Getting a copy of columns to modify
-path_names, starts, ends = bed["query.name"].tolist(), bed["query.start"].tolist(), bed["query.end"].tolist()
-
-## Iterating over lines
-for i in range(len(path_names)):
- ## Getting the name and the range of the current line query
- path, ranges = path_names[i].split(":")
-
- if path in splitted_paths:
- #print("Before :", path_names[i], starts[i], ends[i])
-
- ## Computing offset based on min start for this path
- offset = (int(ranges.split('-')[0])-min_start[path])
- #print("Offset:", offset)
-
- ## Patching the info with new range (min start, max end) and shifted coordinates
- path_names[i] = f"{path}:{min_start[path]}-{max_end[path]}"
- starts[i] = int(starts[i])+offset
- ends[i] = int(ends[i])+offset
- #print("After :", path_names[i], starts[i], ends[i])
-
-## Patching the bed
-bed["query.name"] = path_names
-bed["query.start"] = starts
-bed["query.end"] = ends
-
-# Exporting
-bed.to_csv(args.output, sep="\t", index = False)
\ No newline at end of file
diff --git a/gaf2aln.ipynb b/gaf2aln.ipynb
deleted file mode 100644
index 04fe866..0000000
--- a/gaf2aln.ipynb
+++ /dev/null
@@ -1,2443 +0,0 @@
-{
- "cells": [
- {
- "cell_type": "code",
- "execution_count": 1,
- "id": "4ffaf9f6-cc1e-4190-9351-5431c930d25b",
- "metadata": {},
- "outputs": [],
- "source": [
- "import numpy as np\n",
- "import pandas as pd\n",
- "import argparse\n",
- "import concurrent.futures\n",
- "import os\n",
- "import re\n",
- "\n",
- "# Replace for argparse arguments\n",
- "class arguments():\n",
- " gfa = \"/home/amergez/Documents/Scratch/LeChou/35Bra-v2a/35Bra-v2a.chr03.gfa\"\n",
- " gaf = \"/home/amergez/Documents/Scratch/LeChou/35Bra-v2a/Mapping2Graph/GA.FLC2.aln.gaf\"\n",
- " threads = 8\n",
- " version = False\n",
- "args = arguments()"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 2,
- "id": "280c8847-22e8-4063-bde8-3e4e72cf20e7",
- "metadata": {},
- "outputs": [],
- "source": [
- "# Toolbox\n",
- "def walk2path(walk):\n",
- " \"\"\"\n",
- " Takes a walk in a single string and returns a list of nodes id with signs (gfa v1 like)\n",
- " \"\"\"\n",
- " _ = re.findall(r'>\\w+|<\\w+', walk)\n",
- " # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'\n",
- " return [f'{elem[1:]}{(elem[0] == \">\")*\"+\"+(elem[0] == \"<\")*\"-\"}' for elem in _]\n",
- "\n",
- "def cigar2basealn(cigar):\n",
- " \"\"\"\n",
- " Takes a CIGAR string and convert it into a list of base level alignment.\n",
- " For example : \"345=\" -> [\"=\", \"=\", ..., \"=\"] of length 345.\n",
- " \"\"\"\n",
- " _ = re.findall(r'\\d+\\D', cigar)\n",
- " final_cigar = []\n",
- " for match in _:\n",
- " final_cigar += [match[-1]]*int(match[:-1])\n",
- "\n",
- " return final_cigar\n",
- "\n",
- "def basealn2cigar(base_aln_list):\n",
- " \n",
- " last_elem = base_aln_list[0]\n",
- " CIGAR = [[1, last_elem]]\n",
- " for elem in base_aln_list[1:]:\n",
- " if elem == last_elem:\n",
- " CIGAR[-1][0] += 1\n",
- "\n",
- " else :\n",
- " CIGAR[-1][0] = str(CIGAR[-1][0])\n",
- " CIGAR.append([1, elem])\n",
- " last_elem = elem\n",
- " CIGAR[-1][0] = str(CIGAR[-1][0])\n",
- " return \"\".join([\"\".join(block) for block in CIGAR if block[1] != \"\"])"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 3,
- "id": "be12e9d4-de76-4c8b-af84-6567549483f4",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "[gaf2aln::GAF Parser] Reading /home/amergez/Documents/Scratch/LeChou/35Bra-v2a/Mapping2Graph/GA.FLC2.aln.gaf ...\n",
- "[gaf2aln::GAF Parser] Extracting alignments ...\n",
- "{'ALN_1': {'QRY.NAME': 'FLC2.TO1000#1#chr03', 'QRY.LEN': '3735', 'QRY.START': '0', 'QRY.END': '3735', 'STRAND': '+', 'PATH.MATCH': [('7046526', '+'), ('7046528', '+'), ('7046530', '+'), ('7046531', '+'), ('7046532', '+'), ('7046533', '+'), ('7046534', '+'), ('7046536', '+'), ('7046537', '+'), ('7046539', '+'), ('7046541', '+'), ('7046542', '+'), ('7046544', '+'), ('7046546', '+'), ('7046547', '+'), ('7046549', '+'), ('7046551', '+'), ('7046552', '+'), ('7046554', '+'), ('7046556', '+'), ('7046556', '+'), ('7046556', '+'), ('7046557', '+'), ('7046558', '+'), ('7046559', '+'), ('7046560', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046561', '+'), ('7046562', '+'), ('7046564', '+'), ('7046565', '+'), ('7046567', '+'), ('7046568', '+'), ('7046570', '+'), ('7046571', '+'), ('7046573', '+'), ('7046574', '+'), ('7046576', '+'), ('7046577', '+'), ('7046579', '+'), ('7046581', '+'), ('7046583', '+'), ('7046584', '+'), ('7046586', '+'), ('7046587', '+'), ('7046589', '+'), ('7046590', '+'), ('7046592', '+'), ('7046593', '+'), ('7046594', '+'), ('7046596', '+'), ('7046597', '+'), ('7046599', '+'), ('7046600', '+'), ('7046601', '+'), ('7046603', '+'), ('7046604', '+'), ('7046606', '+'), ('7046608', '+'), ('7046609', '+'), ('7046621', '+'), ('7046622', '+'), ('7046624', '+'), ('7046625', '+'), ('7046626', '+'), ('7046628', '+'), ('7046631', '+'), ('7046673', '+'), ('7046631', '+'), ('7046673', '+'), ('7046631', '+'), ('7046632', '+'), ('7046634', '+'), ('7046635', '+'), ('7046637', '+'), ('7046638', '+'), ('7046639', '+'), ('7046641', '+'), ('7046644', '+'), ('7046646', '+'), ('7046647', '+'), ('7046649', '+'), ('7046650', '+'), ('7046652', '+'), ('7046653', '+'), ('7046654', '+'), ('7046656', '+'), ('7046657', '+'), ('7046659', '+'), ('7046660', '+'), ('7046662', '+'), ('7046663', '+'), ('7046665', '+'), ('7046667', '+'), ('7046668', '+'), ('7046670', '+'), ('7046671', '+'), ('7046674', '+'), ('7046675', '+'), ('7046674', '+'), ('7046675', '+'), ('7046676', '+'), ('7046678', '+'), ('7046679', '+'), ('7046680', '+'), ('7046682', '+'), ('7046684', '+'), ('7046685', '+'), ('7046686', '+'), ('7046688', '+'), ('7046690', '+'), ('7046692', '+'), ('7046693', '+'), ('7046695', '+'), ('7046696', '+'), ('7046698', '+'), ('7046700', '+'), ('7046702', '+'), ('7046703', '+'), ('7046704', '+'), ('7046706', '+'), ('7046707', '+'), ('7046709', '+'), ('7046710', '+'), ('7046712', '+'), ('7046713', '+'), ('7046715', '+'), ('7046718', '+'), ('7046717', '+'), ('7046718', '+'), ('7046717', '+'), ('7046718', '+'), ('7046720', '+'), ('7046722', '+'), ('7046724', '+'), ('7046725', '+'), ('7046727', '+'), ('7046728', '+'), ('7046729', '+'), ('7046730', '+'), ('7046731', '+'), ('7046733', '+'), ('7046735', '+'), ('7046736', '+'), ('7046738', '+'), ('7046739', '+'), ('7046740', '+'), ('7046738', '+'), ('7046739', '+'), ('7046740', '+'), ('7046738', '+'), ('7046739', '+'), ('7046741', '+')], 'PATH.LEN': '3822', 'ALN.START': '77', 'ALN.END': '3812', 'RES.MATCH': '3735', 'ALN.BLOCK.LEN': '3735', 'MAPPING.QUAL': '60', 'RAW.CIGAR': 'cg:Z:3735=', 'TAGS': 'AS:f:3735,dv:f:0,id:f:1'}, 'ALN_2': {'QRY.NAME': 'FLC2.TO1000#1#chr03', 'QRY.LEN': '3735', 'QRY.START': '0', 'QRY.END': '3735', 'STRAND': '+', 'PATH.MATCH': [('7594382', '+'), ('7594369', '+'), ('7594371', '+'), ('7594021', '+'), ('7594286', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594374', '+'), ('7594375', '+'), ('7594626', '+'), ('7594011', '+'), ('7594374', '+'), ('7594375', '+'), ('7594369', '+'), ('7594371', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594241', '+'), ('7594248', '+'), ('7594286', '+'), ('7594311', '+'), ('7594315', '+'), ('7594311', '+'), ('7594330', '+'), ('7594311', '+'), ('7594315', '+'), ('7594374', '+'), ('7594311', '+'), ('7594374', '+'), ('7594369', '+'), ('7594021', '+'), ('7594026', '+'), ('7594021', '+'), ('7594021', '+'), ('7594026', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594021', '+'), ('7594286', '+'), ('7594374', '+'), ('7594021', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594286', '+'), ('7594311', '+'), ('7594315', '+'), ('7594286', '+'), ('7594311', '+'), ('7594374', '+'), ('7594021', '+'), ('7594286', '+'), ('7594286', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594374', '+'), ('7594375', '+'), ('7594374', '+'), ('7594356', '+'), ('7594374', '+'), ('7594375', '+'), ('7594374', '+'), ('7594350', '+'), ('7594264', '+'), ('7594207', '+'), ('7594225', '+'), ('7594227', '+'), ('7594120', '+'), ('7594132', '+'), ('7594165', '+'), ('7594172', '+')], 'PATH.LEN': '61224', 'ALN.START': '0', 'ALN.END': '3735', 'RES.MATCH': '3734', 'ALN.BLOCK.LEN': '3735', 'MAPPING.QUAL': '0', 'RAW.CIGAR': 'cg:Z:57=1X3677=', 'TAGS': 'AS:f:3732.06,dv:f:0.000267738,id:f:0.999732'}}\n"
- ]
- }
- ],
- "source": [
- "# Parsing the .gaf file\n",
- "print(f\"[gaf2aln::GAF Parser] Reading {args.gaf} ...\")\n",
- "with open(args.gaf, 'r') as file:\n",
- " gaf_lines = file.readlines()\n",
- "\n",
- "gaf_col = [\n",
- " \"QRY.NAME\", \"QRY.LEN\", \"QRY.START\", \"QRY.END\", \"STRAND\", \n",
- " \"PATH.MATCH\", \"PATH.LEN\", \"ALN.START\", \"ALN.END\",\n",
- " \"RES.MATCH\", \"ALN.BLOCK.LEN\", \"MAPPING.QUAL\"\n",
- " ]\n",
- "\n",
- "# Creating dictionnary to store alignments\n",
- "print(f\"[gaf2aln::GAF Parser] Extracting alignments ...\")\n",
- "aln_dict = {}\n",
- "for line in range(len(gaf_lines)):\n",
- " ## Splitting the line by tabulation\n",
- " line_content = gaf_lines[line][:-1].split('\\t')\n",
- "\n",
- " ## Adding alignement info to dictionnary\n",
- " aln_dict[f\"ALN_{line+1}\"] = {\n",
- " gaf_col[i]: line_content[i] for i in range(len(gaf_col))\n",
- " }\n",
- " \n",
- " ## Splitting \"PATH.MATCH\" into a list\n",
- " aln_dict[f\"ALN_{line+1}\"][\"PATH.MATCH\"] = [\n",
- " (str(node_id[:-1]), node_id[-1]) \n",
- " for node_id in walk2path(aln_dict[f\"ALN_{line+1}\"][\"PATH.MATCH\"])\n",
- " ]\n",
- "\n",
- " ## Adding CIGAR\n",
- " aln_dict[f\"ALN_{line+1}\"][\"RAW.CIGAR\"] = line_content[-1]\n",
- "\n",
- " ## Adding tags\n",
- " aln_dict[f\"ALN_{line+1}\"][\"TAGS\"] = \",\".join(line_content[13:-1])\n",
- "\n",
- "# Getting nodes of interest ids\n",
- "aln_nodes = np.unique([\n",
- " str(node_id) \n",
- " for aln in aln_dict.keys() \n",
- " for node_id, orient in aln_dict[aln][\"PATH.MATCH\"]\n",
- "]).tolist()\n",
- "\n",
- "print(aln_dict)\n",
- "del gaf_lines, gaf_col"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 4,
- "id": "2f891424-0d88-4fd3-99ff-b0a8c90587ff",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "[gaf2aln::GFA Parser] Reading /home/amergez/Documents/Scratch/LeChou/35Bra-v2a/35Bra-v2a.chr03.gfa ...\n",
- "[gaf2aln::GFA Parser] Extracting nodes, paths and links ...\n"
- ]
- }
- ],
- "source": [
- "# Parsing the .gfa\n",
- "print(f\"[gaf2aln::GFA Parser] Reading {args.gfa} ...\")\n",
- "with open(args.gfa, 'r') as file:\n",
- " gfa_lines = file.readlines()\n",
- "\n",
- "# Nodes length dictionnary structured as follow :\n",
- "# {<NODE.ID>: <NODE.LENGTH>}\n",
- "nodes_length = {}\n",
- "# Nodes dictionnary structured as follow :\n",
- "# { <ALN.NODE.ID> : {\n",
- "# <PATH.NAME>: {\"START\": start, \"END\": end, \"STRAND\": strand), \n",
- "# <ALN.NAME>: {\"START\": start, \"END\": end, \"S.OFF\": start.offset, \"E.OFF\": end.offset, \"STRAND\": strand, \"CIGAR\": CIGAR}\n",
- "# }\n",
- "# }\n",
- "nodes = {node_id: {} for node_id in aln_nodes}\n",
- "# Paths dictionnary structured as follow :\n",
- "# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}\n",
- "paths = {}\n",
- "# Links dictionnary structured as follow : \n",
- "# {<FROM.NODE.ID>: {<TO.NODE.ID>: {FROM.ORIENT: <FROM.ORIENT>, TO.ORIENT: <TO.ORIENT>}}}\n",
- "links = {}\n",
- "\n",
- "# Parsing the gfa\n",
- "print(f\"[gaf2aln::GFA Parser] Extracting nodes, paths and links ...\")\n",
- "\n",
- "def GFA_parser(gfa_lines, nodes = nodes):\n",
- " _links, _nodes, _nodes_length, paths = {}, {}, {}, {}\n",
- " for line in gfa_lines:\n",
- " line_content = line[:-1].split(\"\\t\")\n",
- " line_id = line_content[0]\n",
- " \n",
- " # Segment line\n",
- " if line_id == \"S\" :\n",
- " \n",
- " _nodes_length[str(line_content[1])] = len(line_content[2])\n",
- " \n",
- " # Link line\n",
- " elif line_id == \"L\":\n",
- " try :\n",
- " _links[str(line_content[1])][str(line_content[3])] = {\n",
- " \"FROM\": str(line_content[2]), \n",
- " \"TO\": str(line_content[4])\n",
- " }\n",
- "\n",
- " except :\n",
- " _links[str(line_content[1])] = {\n",
- " str(line_content[3]) : {\"FROM.ORIENT\": str(line_content[2]), \"TO.ORIENT\": str(line_content[4])}\n",
- " }\n",
- "\n",
- " # Path line\n",
- " elif line_id == \"P\":\n",
- " _paths[str(line_content[1])] = {\n",
- " \"NODES\": {\n",
- " str(node_id[:-1]): str(node_id[-1])\n",
- " for node_id in line_content[2].split(',')\n",
- " },\n",
- " \"CIGAR\": line_content[3]\n",
- " }\n",
- "\n",
- " return [_links, _nodes, _nodes_length, _paths]\n",
- "\n",
- "# splits = np.quantile(range(len(gfa_lines)+1), q= np.array(args.threads+1)/args.threads, method='higher').tolist()\n",
- "# res = []\n",
- "# for i in range(1, len(splits)):\n",
- "# res.append(executor.submit(GFA_parser, gfa_lines[splits[i-1]:splits[i]]))\n",
- "\n",
- "# for out in res:\n",
- "# results = out.result()\n",
- "\n",
- "# for link_id, link_info in results[0].items():\n",
- "# links[]\n",
- "\n",
- "\n",
- "for line in gfa_lines:\n",
- " line_content = line[:-1].split(\"\\t\")\n",
- " line_id = line_content[0]\n",
- " \n",
- " # Segment line\n",
- " if line_id == \"S\" :\n",
- " \n",
- " nodes_length[str(line_content[1])] = len(line_content[2])\n",
- " \n",
- " # Link line\n",
- " elif line_id == \"L\":\n",
- " try :\n",
- " links[str(line_content[1])][str(line_content[3])] = {\n",
- " \"FROM\": str(line_content[2]), \n",
- " \"TO\": str(line_content[4])\n",
- " }\n",
- "\n",
- " except :\n",
- " links[str(line_content[1])] = {\n",
- " str(line_content[3]) : {\"FROM.ORIENT\": str(line_content[2]), \"TO.ORIENT\": str(line_content[4])}\n",
- " }\n",
- "\n",
- " # Path line\n",
- " elif line_id == \"P\":\n",
- " paths[str(line_content[1])] = {\n",
- " \"NODES\": {\n",
- " str(node_id[:-1]): str(node_id[-1])\n",
- " for node_id in line_content[2].split(',')\n",
- " },\n",
- " \"CIGAR\": line_content[3]\n",
- " }\n",
- "\n",
- "del gfa_lines"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 5,
- "id": "a403c88e-54ea-4a67-9047-dc44eba7f51a",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "[gaf2aln::Graph position processing] Computing nodes positions in each paths...\n",
- "[gaf2aln::Graph position processing] Running on Capitata#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on D101#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on D134#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on G06-09-28#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on G07-DH-33#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on HDEM#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on Korso#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on M249#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on OX-heart#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on PL021#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on RC34#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T02#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T03#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T04#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T06#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T07#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T08#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T09#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T10#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T11#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T12#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T13#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T14#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T15#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T16#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T17#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T18#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T19#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T21#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T24#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T25#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T26#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on T27#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on TO1000#1#chr03 ...\n",
- "[gaf2aln::Graph position processing] Running on W1701#1#chr03 ...\n"
- ]
- }
- ],
- "source": [
- "print(f\"[gaf2aln::Graph position processing] Computing nodes positions in each paths...\")\n",
- "def get_node_pos(path_name, nodes = nodes, paths = paths, nodes_length = nodes_length):\n",
- " print(f\"[gaf2aln::Graph position processing] Running on {path_name} ...\")\n",
- " cur_pos = 0\n",
- "\n",
- " out = {}\n",
- " # Iterating over nodes in the path\n",
- " for path_node in paths[path_name][\"NODES\"].keys():\n",
- " # Instead of checking if the node is one interesting node, we try to add to the nodes dict\n",
- " if path_node in aln_nodes :\n",
- " out[path_node] = {\n",
- " \"START\": cur_pos, # Start position of the node start in the currrent path\n",
- " \"END\": cur_pos+nodes_length[path_node], # End position of the node end in the current path\n",
- " \"STRAND\": paths[path_name][\"NODES\"][path_node] # Orientation of the node in the current path\n",
- " } \n",
- "\n",
- " cur_pos += nodes_length[path_node]+1\n",
- " else :\n",
- " cur_pos += nodes_length[path_node]+1\n",
- "\n",
- " return out\n",
- "\n",
- "res = {}\n",
- "executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)\n",
- "# Adding nodes positions relative to path\n",
- "for path_name in paths.keys():\n",
- " res[path_name] = executor.submit(get_node_pos, path_name)\n",
- "\n",
- "executor.shutdown(wait=True)\n",
- "\n",
- "for path_name, out in res.items():\n",
- " results = out.result()\n",
- " for path_node, node_pos in results.items():\n",
- " nodes[path_node][path_name] = node_pos\n",
- "\n",
- "del res"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 6,
- "id": "bed36bd5-30eb-4d02-8b52-1ae5d753f8f8",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...\n",
- "[gaf2aln::Alignment position processing] Running on ALN_1 ...\n",
- "0 77 77 0 + 154 77\n",
- "77 82 0 0 + 5 82\n",
- "82 83 0 0 + 1 83\n",
- "83 138 0 0 + 55 138\n",
- "138 139 0 0 + 1 139\n",
- "139 202 0 0 + 63 202\n",
- "202 203 0 0 + 1 203\n",
- "203 379 0 0 + 176 379\n",
- "379 380 0 0 + 1 380\n",
- "380 429 0 0 + 49 429\n",
- "429 430 0 0 + 1 430\n",
- "430 457 0 0 + 27 457\n",
- "457 492 0 0 + 35 492\n",
- "492 494 0 0 + 2 494\n",
- "494 497 0 0 + 3 497\n",
- "497 507 0 0 + 10 507\n",
- "507 508 0 0 + 1 508\n",
- "508 564 0 0 + 56 564\n",
- "564 566 0 0 + 2 566\n",
- "566 567 0 0 + 1 567\n",
- "567 568 0 0 + 1 568\n",
- "568 569 0 0 + 1 569\n",
- "569 824 0 0 + 255 824\n",
- "824 826 0 0 + 2 826\n",
- "826 858 0 0 + 32 858\n",
- "858 859 0 0 + 1 859\n",
- "859 860 0 0 + 1 860\n",
- "860 861 0 0 + 1 861\n",
- "861 862 0 0 + 1 862\n",
- "862 863 0 0 + 1 863\n",
- "863 864 0 0 + 1 864\n",
- "864 865 0 0 + 1 865\n",
- "865 866 0 0 + 1 866\n",
- "866 867 0 0 + 1 867\n",
- "867 868 0 0 + 1 868\n",
- "868 869 0 0 + 1 869\n",
- "869 913 0 0 + 44 913\n",
- "913 919 0 0 + 6 919\n",
- "919 978 0 0 + 59 978\n",
- "978 979 0 0 + 1 979\n",
- "979 1038 0 0 + 59 1038\n",
- "1038 1045 0 0 + 7 1045\n",
- "1045 1046 0 0 + 1 1046\n",
- "1046 1080 0 0 + 34 1080\n",
- "1080 1081 0 0 + 1 1081\n",
- "1081 1107 0 0 + 26 1107\n",
- "1107 1108 0 0 + 1 1108\n",
- "1108 1183 0 0 + 75 1183\n",
- "1183 1186 0 0 + 3 1186\n",
- "1186 1224 0 0 + 38 1224\n",
- "1224 1257 0 0 + 33 1257\n",
- "1257 1289 0 0 + 32 1289\n",
- "1289 1311 0 0 + 22 1311\n",
- "1311 1359 0 0 + 48 1359\n",
- "1359 1382 0 0 + 23 1382\n",
- "1382 1434 0 0 + 52 1434\n",
- "1434 1451 0 0 + 17 1451\n",
- "1451 1531 0 0 + 80 1531\n",
- "1531 1532 0 0 + 1 1532\n",
- "1532 1543 0 0 + 11 1543\n",
- "1543 1544 0 0 + 1 1544\n",
- "1544 1572 0 0 + 28 1572\n",
- "1572 1573 0 0 + 1 1573\n",
- "1573 1587 0 0 + 14 1587\n",
- "1587 1588 0 0 + 1 1588\n",
- "1588 1616 0 0 + 28 1616\n",
- "1616 1617 0 0 + 1 1617\n",
- "1617 1646 0 0 + 29 1646\n",
- "1646 1661 0 0 + 15 1661\n",
- "1661 1673 0 0 + 12 1673\n",
- "1673 1674 0 0 + 1 1674\n",
- "1674 1726 0 0 + 52 1726\n",
- "1726 1727 0 0 + 1 1727\n",
- "1727 1762 0 0 + 35 1762\n",
- "1762 1763 0 0 + 1 1763\n",
- "1763 1764 0 0 + 1 1764\n",
- "1764 1765 0 0 + 1 1765\n",
- "1765 1766 0 0 + 1 1766\n",
- "1766 1767 0 0 + 1 1767\n",
- "1767 1824 0 0 + 57 1824\n",
- "1824 1825 0 0 + 1 1825\n",
- "1825 1975 0 0 + 150 1975\n",
- "1975 1976 0 0 + 1 1976\n",
- "1976 2015 0 0 + 39 2015\n",
- "2015 2016 0 0 + 1 2016\n",
- "2016 2047 0 0 + 31 2047\n",
- "2047 2055 0 0 + 8 2055\n",
- "2055 2056 0 0 + 1 2056\n",
- "2056 2120 0 0 + 64 2120\n",
- "2120 2121 0 0 + 1 2121\n",
- "2121 2157 0 0 + 36 2157\n",
- "2157 2158 0 0 + 1 2158\n",
- "2158 2170 0 0 + 12 2170\n",
- "2170 2171 0 0 + 1 2171\n",
- "2171 2205 0 0 + 34 2205\n",
- "2205 2206 0 0 + 1 2206\n",
- "2206 2344 0 0 + 138 2344\n",
- "2344 2345 0 0 + 1 2345\n",
- "2345 2364 0 0 + 19 2364\n",
- "2364 2383 0 0 + 19 2383\n",
- "2383 2408 0 0 + 25 2408\n",
- "2408 2409 0 0 + 1 2409\n",
- "2409 2441 0 0 + 32 2441\n",
- "2441 2442 0 0 + 1 2442\n",
- "2442 2580 0 0 + 138 2580\n",
- "2580 2581 0 0 + 1 2581\n",
- "2581 2582 0 0 + 1 2582\n",
- "2582 2583 0 0 + 1 2583\n",
- "2583 2584 0 0 + 1 2584\n",
- "2584 2764 0 0 + 180 2764\n",
- "2764 2765 0 0 + 1 2765\n",
- "2765 2797 0 0 + 32 2797\n",
- "2797 2798 0 0 + 1 2798\n",
- "2798 2878 0 0 + 80 2878\n",
- "2878 2879 0 0 + 1 2879\n",
- "2879 2951 0 0 + 72 2951\n",
- "2951 2952 0 0 + 1 2952\n",
- "2952 3002 0 0 + 50 3002\n",
- "3002 3077 0 0 + 75 3077\n",
- "3077 3078 0 0 + 1 3078\n",
- "3078 3093 0 0 + 15 3093\n",
- "3093 3094 0 0 + 1 3094\n",
- "3094 3097 0 0 + 3 3097\n",
- "3097 3140 0 0 + 43 3140\n",
- "3140 3210 0 0 + 70 3210\n",
- "3210 3211 0 0 + 1 3211\n",
- "3211 3229 0 0 + 18 3229\n",
- "3229 3230 0 0 + 1 3230\n",
- "3230 3276 0 0 + 46 3276\n",
- "3276 3277 0 0 + 1 3277\n",
- "3277 3315 0 0 + 38 3315\n",
- "3315 3316 0 0 + 1 3316\n",
- "3316 3322 0 0 + 6 3322\n",
- "3322 3323 0 0 + 1 3323\n",
- "3323 3348 0 0 + 25 3348\n",
- "3348 3349 0 0 + 1 3349\n",
- "3349 3350 0 0 + 1 3350\n",
- "3350 3351 0 0 + 1 3351\n",
- "3351 3352 0 0 + 1 3352\n",
- "3352 3353 0 0 + 1 3353\n",
- "3353 3354 0 0 + 1 3354\n",
- "3354 3356 0 0 + 2 3356\n",
- "3356 3357 0 0 + 1 3357\n",
- "3357 3489 0 0 + 132 3489\n",
- "3489 3490 0 0 + 1 3490\n",
- "3490 3642 0 0 + 152 3642\n",
- "3642 3644 0 0 + 2 3644\n",
- "3644 3685 0 0 + 41 3685\n",
- "3685 3687 0 0 + 2 3687\n",
- "3687 3693 0 0 + 6 3693\n",
- "3693 3694 0 0 + 1 3694\n",
- "3694 3708 0 0 + 14 3708\n",
- "3708 3709 0 0 + 1 3709\n",
- "3709 3710 0 0 + 1 3710\n",
- "3710 3714 0 0 + 4 3714\n",
- "3714 3715 0 0 + 1 3715\n",
- "3715 3716 0 0 + 1 3716\n",
- "3716 3720 0 0 + 4 3720\n",
- "3720 3721 0 0 + 1 3721\n",
- "3721 3722 0 0 + 1 3722\n",
- "3722 3735 0 10 + 23 3735\n",
- "[gaf2aln::Alignment position processing] Running on ALN_2 ...\n",
- "0 1 0 0 + 1 1\n",
- "1 2 0 0 + 1 2\n",
- "2 3 0 0 + 1 3\n",
- "3 4 0 0 + 1 4\n",
- "4 5 0 0 + 1 5\n",
- "5 6 0 0 + 1 6\n",
- "6 7 0 0 + 1 7\n",
- "7 8 0 0 + 1 8\n",
- "8 9 0 0 + 1 9\n",
- "9 10 0 0 + 1 10\n",
- "10 11 0 0 + 1 11\n",
- "11 12 0 0 + 1 12\n",
- "12 13 0 0 + 1 13\n",
- "13 14 0 0 + 1 14\n",
- "14 15 0 0 + 1 15\n",
- "15 16 0 0 + 1 16\n",
- "16 17 0 0 + 1 17\n",
- "17 18 0 0 + 1 18\n",
- "18 19 0 0 + 1 19\n",
- "19 20 0 0 + 1 20\n",
- "20 21 0 0 + 1 21\n",
- "21 22 0 0 + 1 22\n",
- "22 23 0 0 + 1 23\n",
- "23 24 0 0 + 1 24\n",
- "24 25 0 0 + 1 25\n",
- "25 26 0 0 + 1 26\n",
- "26 27 0 0 + 1 27\n",
- "27 28 0 0 + 1 28\n",
- "28 29 0 0 + 1 29\n",
- "29 30 0 0 + 1 30\n",
- "30 31 0 0 + 1 31\n",
- "31 32 0 0 + 1 32\n",
- "32 33 0 0 + 1 33\n",
- "33 34 0 0 + 1 34\n",
- "34 35 0 0 + 1 35\n",
- "35 36 0 0 + 1 36\n",
- "36 37 0 0 + 1 37\n",
- "37 38 0 0 + 1 38\n",
- "38 39 0 0 + 1 39\n",
- "39 40 0 0 + 1 40\n",
- "40 41 0 0 + 1 41\n",
- "41 42 0 0 + 1 42\n",
- "42 43 0 0 + 1 43\n",
- "43 44 0 0 + 1 44\n",
- "44 45 0 0 + 1 45\n",
- "45 46 0 0 + 1 46\n",
- "46 47 0 0 + 1 47\n",
- "47 48 0 0 + 1 48\n",
- "48 49 0 0 + 1 49\n",
- "49 50 0 0 + 1 50\n",
- "50 51 0 0 + 1 51\n",
- "51 52 0 0 + 1 52\n",
- "52 53 0 0 + 1 53\n",
- "53 54 0 0 + 1 54\n",
- "54 55 0 0 + 1 55\n",
- "55 56 0 0 + 1 56\n",
- "56 57 0 0 + 1 57\n",
- "57 58 0 0 + 1 58\n",
- "58 59 0 0 + 1 59\n",
- "59 60 0 0 + 1 60\n",
- "60 61 0 0 + 1 61\n",
- "61 62 0 0 + 1 62\n",
- "62 63 0 0 + 1 63\n",
- "63 64 0 0 + 1 64\n",
- "64 65 0 0 + 1 65\n",
- "65 66 0 0 + 1 66\n",
- "66 67 0 0 + 1 67\n",
- "67 68 0 0 + 1 68\n",
- "68 69 0 0 + 1 69\n",
- "69 70 0 0 + 1 70\n",
- "70 71 0 0 + 1 71\n",
- "71 72 0 0 + 1 72\n",
- "72 73 0 0 + 1 73\n",
- "73 74 0 0 + 1 74\n",
- "74 75 0 0 + 1 75\n",
- "75 76 0 0 + 1 76\n",
- "76 77 0 0 + 1 77\n",
- "77 78 0 0 + 1 78\n",
- "78 3735 0 57489 + 61146 3735\n"
- ]
- }
- ],
- "source": [
- "print(f\"[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...\")\n",
- "# Adding nodes positions relative to path\n",
- "\n",
- "def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length):\n",
- " # Initializing current position in query\n",
- " cur_pos = 0\n",
- "\n",
- " # Getting start and end node ids\n",
- " start_end_id = (aln_dict[aln_name][\"PATH.MATCH\"][0][0], aln_dict[aln_name][\"PATH.MATCH\"][-1][0])\n",
- "\n",
- " # Creating result dictionnary\n",
- " res = {}\n",
- "\n",
- " ## Iterating over node_ids from the given alignment\n",
- " for node_id, orient in aln_dict[aln_name][\"PATH.MATCH\"]:\n",
- " # Adding entry for current node\n",
- " res[node_id] = {aln_name: {}}\n",
- "\n",
- " # First node\n",
- " if node_id == start_end_id[0]:\n",
- " start_pos = 0\n",
- " s_off = int(aln_dict[aln_name][\"ALN.START\"])\n",
- " end_pos = nodes_length[node_id]-s_off\n",
- " e_off = 0\n",
- " # End node\n",
- " elif node_id == start_end_id[1]:\n",
- " start_pos = cur_pos\n",
- " s_off = 0\n",
- " end_pos = int(aln_dict[aln_name][\"QRY.END\"])\n",
- " e_off = nodes_length[node_id]-(end_pos-cur_pos)\n",
- " # Node in between\n",
- " else :\n",
- " start_pos = cur_pos\n",
- " s_off, e_off = 0, 0\n",
- " end_pos = cur_pos+nodes_length[node_id]\n",
- "\n",
- " res[node_id] = {\n",
- " \"START\": start_pos, # Start position on the query\n",
- " \"END\": end_pos, # End position on the query\n",
- " \"S.OFF\": s_off, # Offset between the start of the alignment and the node's start\n",
- " \"E.OFF\": e_off, # Offset between the end of the alignment and the node's end \n",
- " \"STRAND\": orient # Orientation of the node in the alignment\n",
- " }\n",
- " \n",
- " cur_pos = end_pos\n",
- " print(start_pos, end_pos, s_off, e_off, orient, nodes_length[node_id], cur_pos)\n",
- "\n",
- " return res\n",
- "\n",
- "# Storing alignement \n",
- "res = {}\n",
- "executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)\n",
- "for aln_name in aln_dict.keys():\n",
- " print(f\"[gaf2aln::Alignment position processing] Running on {aln_name} ...\")\n",
- " \n",
- " res[aln_name] = executor.submit(get_aln_node_info, aln_name)\n",
- " #res[aln_name] = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)\n",
- "\n",
- "executor.shutdown(wait=True)\n",
- "\n",
- "for aln_name, node_info in res.items():\n",
- " results = node_info.result()\n",
- " for node_id, info in results.items():\n",
- " nodes[node_id][aln_name] = info\n",
- "\n",
- "del res"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 7,
- "id": "4c30727c-7ffc-4852-ad81-ca2a5a7f9957",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...\n",
- "[gaf2aln::CIGAR processing] Running on ALN_1 ...\n",
- "[gaf2aln::CIGAR processing] Running on ALN_2 ...\n"
- ]
- }
- ],
- "source": [
- "# Calculating CIGAR for each nodes in each aln\n",
- "print(f\"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...\")\n",
- "# Iterating over alignments\n",
- "for aln in aln_dict.keys():\n",
- " \n",
- " print(f\"[gaf2aln::CIGAR processing] Running on {aln} ...\")\n",
- " # Getting the list of base level alignement ([\"=\", \"X\", ...] from \"1=1X...\")\n",
- " raw_cigar = cigar2basealn(aln_dict[aln][\"RAW.CIGAR\"])\n",
- " CIGAR={}\n",
- "\n",
- " for node_id, orient in aln_dict[aln][\"PATH.MATCH\"]:\n",
- "\n",
- " _cigar = basealn2cigar(raw_cigar[\n",
- " nodes[node_id][aln][\"START\"]:nodes[node_id][aln][\"END\"]\n",
- " ])\n",
- " nodes[node_id][aln][\"CIGAR\"] = _cigar\n",
- " #print(_cigar, nodes[node_id][aln][\"START\"], nodes[node_id][aln][\"END\"])"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 8,
- "id": "e15e4762-cd71-4afe-bc74-ebe44869fee6",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "ALN_1\n",
- "7046526 D134#1#chr03 0 77\n",
- "\tIn path\n",
- "\t 73306158 73306235\n",
- "skipped\n",
- "\n",
- "7046528 D134#1#chr03 77 82\n",
- "\tIn path\n",
- "\t 73306238 73306243\n",
- "{'Q.START': 0, 'Q.END': 77, 'T.START': 73306158, 'T.END': 73306235, 'CG': '77='}\n",
- "7046530 D134#1#chr03 82 83\n",
- "\tNot in path\n",
- "7046531 D134#1#chr03 83 138\n",
- "\tIn path\n",
- "\t 73306246 73306301\n",
- "{'Q.START': 77, 'Q.END': 82, 'T.START': 73306238, 'T.END': 73306243, 'CG': '5='}\n",
- "7046532 D134#1#chr03 138 139\n",
- "\tNot in path\n",
- "7046533 D134#1#chr03 139 202\n",
- "\tIn path\n",
- "\t 73306302 73306365\n",
- "{'Q.START': 83, 'Q.END': 138, 'T.START': 73306246, 'T.END': 73306301, 'CG': '55='}\n",
- "7046534 D134#1#chr03 202 203\n",
- "\tIn path\n",
- "\t 73306366 73306367\n",
- "{'Q.START': 139, 'Q.END': 202, 'T.START': 73306302, 'T.END': 73306365, 'CG': '63='}\n",
- "7046536 D134#1#chr03 203 379\n",
- "\tIn path\n",
- "\t 73306368 73306544\n",
- "{'Q.START': 202, 'Q.END': 203, 'T.START': 73306366, 'T.END': 73306367, 'CG': '1='}\n",
- "7046537 D134#1#chr03 379 380\n",
- "\tIn path\n",
- "\t 73306545 73306546\n",
- "{'Q.START': 203, 'Q.END': 379, 'T.START': 73306368, 'T.END': 73306544, 'CG': '176='}\n",
- "7046539 D134#1#chr03 380 429\n",
- "\tIn path\n",
- "\t 73306547 73306596\n",
- "{'Q.START': 379, 'Q.END': 380, 'T.START': 73306545, 'T.END': 73306546, 'CG': '1='}\n",
- "7046541 D134#1#chr03 429 430\n",
- "\tIn path\n",
- "\t 73306597 73306598\n",
- "{'Q.START': 380, 'Q.END': 429, 'T.START': 73306547, 'T.END': 73306596, 'CG': '49='}\n",
- "7046542 D134#1#chr03 430 457\n",
- "\tIn path\n",
- "\t 73306599 73306626\n",
- "{'Q.START': 429, 'Q.END': 430, 'T.START': 73306597, 'T.END': 73306598, 'CG': '1='}\n",
- "7046544 D134#1#chr03 457 492\n",
- "\tIn path\n",
- "\t 73306641 73306676\n",
- "{'Q.START': 430, 'Q.END': 457, 'T.START': 73306599, 'T.END': 73306626, 'CG': '27='}\n",
- "7046546 D134#1#chr03 492 494\n",
- "\tNot in path\n",
- "7046547 D134#1#chr03 494 497\n",
- "\tNot in path\n",
- "7046549 D134#1#chr03 497 507\n",
- "\tNot in path\n",
- "7046551 D134#1#chr03 507 508\n",
- "\tNot in path\n",
- "7046552 D134#1#chr03 508 564\n",
- "\tIn path\n",
- "\t 73306694 73306750\n",
- "{'Q.START': 457, 'Q.END': 492, 'T.START': 73306641, 'T.END': 73306676, 'CG': '35='}\n",
- "7046554 D134#1#chr03 564 566\n",
- "\tNot in path\n",
- "7046556 D134#1#chr03 568 569\n",
- "\tIn path\n",
- "\t 73306753 73306754\n",
- "{'Q.START': 508, 'Q.END': 564, 'T.START': 73306694, 'T.END': 73306750, 'CG': '56='}\n",
- "7046556 D134#1#chr03 568 569\n",
- "\tIn path\n",
- "\t 73306753 73306754\n",
- "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
- "7046556 D134#1#chr03 568 569\n",
- "\tIn path\n",
- "\t 73306753 73306754\n",
- "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
- "7046557 D134#1#chr03 569 824\n",
- "\tIn path\n",
- "\t 73306755 73307010\n",
- "{'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}\n",
- "7046558 D134#1#chr03 824 826\n",
- "\tNot in path\n",
- "7046559 D134#1#chr03 826 858\n",
- "\tIn path\n",
- "\t 73307011 73307043\n",
- "{'Q.START': 569, 'Q.END': 824, 'T.START': 73306755, 'T.END': 73307010, 'CG': '255='}\n",
- "7046560 D134#1#chr03 858 859\n",
- "\tIn path\n",
- "\t 73307044 73307045\n",
- "{'Q.START': 826, 'Q.END': 858, 'T.START': 73307011, 'T.END': 73307043, 'CG': '32='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 858, 'Q.END': 859, 'T.START': 73307044, 'T.END': 73307045, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046561 D134#1#chr03 868 869\n",
- "\tIn path\n",
- "\t 73307046 73307047\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046562 D134#1#chr03 869 913\n",
- "\tIn path\n",
- "\t 73307048 73307092\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}\n",
- "7046564 D134#1#chr03 913 919\n",
- "\tIn path\n",
- "\t 73307093 73307099\n",
- "{'Q.START': 869, 'Q.END': 913, 'T.START': 73307048, 'T.END': 73307092, 'CG': '44='}\n",
- "7046565 D134#1#chr03 919 978\n",
- "\tIn path\n",
- "\t 73307100 73307159\n",
- "{'Q.START': 913, 'Q.END': 919, 'T.START': 73307093, 'T.END': 73307099, 'CG': '6='}\n",
- "7046567 D134#1#chr03 978 979\n",
- "\tIn path\n",
- "\t 73307160 73307161\n",
- "{'Q.START': 919, 'Q.END': 978, 'T.START': 73307100, 'T.END': 73307159, 'CG': '59='}\n",
- "7046568 D134#1#chr03 979 1038\n",
- "\tIn path\n",
- "\t 73307162 73307221\n",
- "{'Q.START': 978, 'Q.END': 979, 'T.START': 73307160, 'T.END': 73307161, 'CG': '1='}\n",
- "7046570 D134#1#chr03 1038 1045\n",
- "\tIn path\n",
- "\t 73307224 73307231\n",
- "{'Q.START': 979, 'Q.END': 1038, 'T.START': 73307162, 'T.END': 73307221, 'CG': '59='}\n",
- "7046571 D134#1#chr03 1045 1046\n",
- "\tIn path\n",
- "\t 73307232 73307233\n",
- "{'Q.START': 1038, 'Q.END': 1045, 'T.START': 73307224, 'T.END': 73307231, 'CG': '7='}\n",
- "7046573 D134#1#chr03 1046 1080\n",
- "\tIn path\n",
- "\t 73307234 73307268\n",
- "{'Q.START': 1045, 'Q.END': 1046, 'T.START': 73307232, 'T.END': 73307233, 'CG': '1='}\n",
- "7046574 D134#1#chr03 1080 1081\n",
- "\tIn path\n",
- "\t 73307269 73307270\n",
- "{'Q.START': 1046, 'Q.END': 1080, 'T.START': 73307234, 'T.END': 73307268, 'CG': '34='}\n",
- "7046576 D134#1#chr03 1081 1107\n",
- "\tIn path\n",
- "\t 73307271 73307297\n",
- "{'Q.START': 1080, 'Q.END': 1081, 'T.START': 73307269, 'T.END': 73307270, 'CG': '1='}\n",
- "7046577 D134#1#chr03 1107 1108\n",
- "\tNot in path\n",
- "7046579 D134#1#chr03 1108 1183\n",
- "\tIn path\n",
- "\t 73307300 73307375\n",
- "{'Q.START': 1081, 'Q.END': 1107, 'T.START': 73307271, 'T.END': 73307297, 'CG': '26='}\n",
- "7046581 D134#1#chr03 1183 1186\n",
- "\tIn path\n",
- "\t 73307376 73307379\n",
- "{'Q.START': 1108, 'Q.END': 1183, 'T.START': 73307300, 'T.END': 73307375, 'CG': '75='}\n",
- "7046583 D134#1#chr03 1186 1224\n",
- "\tNot in path\n",
- "7046584 D134#1#chr03 1224 1257\n",
- "\tIn path\n",
- "\t 73307419 73307452\n",
- "{'Q.START': 1183, 'Q.END': 1186, 'T.START': 73307376, 'T.END': 73307379, 'CG': '3='}\n",
- "7046586 D134#1#chr03 1257 1289\n",
- "\tNot in path\n",
- "7046587 D134#1#chr03 1289 1311\n",
- "\tIn path\n",
- "\t 73307475 73307497\n",
- "{'Q.START': 1224, 'Q.END': 1257, 'T.START': 73307419, 'T.END': 73307452, 'CG': '33='}\n",
- "7046589 D134#1#chr03 1311 1359\n",
- "\tNot in path\n",
- "7046590 D134#1#chr03 1359 1382\n",
- "\tIn path\n",
- "\t 73307546 73307569\n",
- "{'Q.START': 1289, 'Q.END': 1311, 'T.START': 73307475, 'T.END': 73307497, 'CG': '22='}\n",
- "7046592 D134#1#chr03 1382 1434\n",
- "\tNot in path\n",
- "7046593 D134#1#chr03 1434 1451\n",
- "\tIn path\n",
- "\t 73307643 73307660\n",
- "{'Q.START': 1359, 'Q.END': 1382, 'T.START': 73307546, 'T.END': 73307569, 'CG': '23='}\n",
- "7046594 D134#1#chr03 1451 1531\n",
- "\tIn path\n",
- "\t 73307661 73307741\n",
- "{'Q.START': 1434, 'Q.END': 1451, 'T.START': 73307643, 'T.END': 73307660, 'CG': '17='}\n",
- "7046596 D134#1#chr03 1531 1532\n",
- "\tNot in path\n",
- "7046597 D134#1#chr03 1532 1543\n",
- "\tIn path\n",
- "\t 73307744 73307755\n",
- "{'Q.START': 1451, 'Q.END': 1531, 'T.START': 73307661, 'T.END': 73307741, 'CG': '80='}\n",
- "7046599 D134#1#chr03 1543 1544\n",
- "\tNot in path\n",
- "7046600 D134#1#chr03 1544 1572\n",
- "\tIn path\n",
- "\t 73307758 73307786\n",
- "{'Q.START': 1532, 'Q.END': 1543, 'T.START': 73307744, 'T.END': 73307755, 'CG': '11='}\n",
- "7046601 D134#1#chr03 1572 1573\n",
- "\tIn path\n",
- "\t 73307787 73307788\n",
- "{'Q.START': 1544, 'Q.END': 1572, 'T.START': 73307758, 'T.END': 73307786, 'CG': '28='}\n",
- "7046603 D134#1#chr03 1573 1587\n",
- "\tIn path\n",
- "\t 73307789 73307803\n",
- "{'Q.START': 1572, 'Q.END': 1573, 'T.START': 73307787, 'T.END': 73307788, 'CG': '1='}\n",
- "7046604 D134#1#chr03 1587 1588\n",
- "\tNot in path\n",
- "7046606 D134#1#chr03 1588 1616\n",
- "\tIn path\n",
- "\t 73307806 73307834\n",
- "{'Q.START': 1573, 'Q.END': 1587, 'T.START': 73307789, 'T.END': 73307803, 'CG': '14='}\n",
- "7046608 D134#1#chr03 1616 1617\n",
- "\tIn path\n",
- "\t 73307835 73307836\n",
- "{'Q.START': 1588, 'Q.END': 1616, 'T.START': 73307806, 'T.END': 73307834, 'CG': '28='}\n",
- "7046609 D134#1#chr03 1617 1646\n",
- "\tIn path\n",
- "\t 73307837 73307866\n",
- "{'Q.START': 1616, 'Q.END': 1617, 'T.START': 73307835, 'T.END': 73307836, 'CG': '1='}\n",
- "7046621 D134#1#chr03 1646 1661\n",
- "\tIn path\n",
- "\t 73307867 73307882\n",
- "{'Q.START': 1617, 'Q.END': 1646, 'T.START': 73307837, 'T.END': 73307866, 'CG': '29='}\n",
- "7046622 D134#1#chr03 1661 1673\n",
- "\tIn path\n",
- "\t 73307883 73307895\n",
- "{'Q.START': 1646, 'Q.END': 1661, 'T.START': 73307867, 'T.END': 73307882, 'CG': '15='}\n",
- "7046624 D134#1#chr03 1673 1674\n",
- "\tIn path\n",
- "\t 73307896 73307897\n",
- "{'Q.START': 1661, 'Q.END': 1673, 'T.START': 73307883, 'T.END': 73307895, 'CG': '12='}\n",
- "7046625 D134#1#chr03 1674 1726\n",
- "\tIn path\n",
- "\t 73307898 73307950\n",
- "{'Q.START': 1673, 'Q.END': 1674, 'T.START': 73307896, 'T.END': 73307897, 'CG': '1='}\n",
- "7046626 D134#1#chr03 1726 1727\n",
- "\tNot in path\n",
- "7046628 D134#1#chr03 1727 1762\n",
- "\tIn path\n",
- "\t 73307953 73307988\n",
- "{'Q.START': 1674, 'Q.END': 1726, 'T.START': 73307898, 'T.END': 73307950, 'CG': '52='}\n",
- "7046631 D134#1#chr03 1766 1767\n",
- "\tIn path\n",
- "\t 73307991 73307992\n",
- "{'Q.START': 1727, 'Q.END': 1762, 'T.START': 73307953, 'T.END': 73307988, 'CG': '35='}\n",
- "7046673 D134#1#chr03 1765 1766\n",
- "\tIn path\n",
- "\t 73307993 73307994\n",
- "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
- "7046631 D134#1#chr03 1766 1767\n",
- "\tIn path\n",
- "\t 73307991 73307992\n",
- "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}\n",
- "7046673 D134#1#chr03 1765 1766\n",
- "\tIn path\n",
- "\t 73307993 73307994\n",
- "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
- "7046631 D134#1#chr03 1766 1767\n",
- "\tIn path\n",
- "\t 73307991 73307992\n",
- "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}\n",
- "7046632 D134#1#chr03 1767 1824\n",
- "\tIn path\n",
- "\t 73307995 73308052\n",
- "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}\n",
- "7046634 D134#1#chr03 1824 1825\n",
- "\tIn path\n",
- "\t 73308053 73308054\n",
- "{'Q.START': 1767, 'Q.END': 1824, 'T.START': 73307995, 'T.END': 73308052, 'CG': '57='}\n",
- "7046635 D134#1#chr03 1825 1975\n",
- "\tIn path\n",
- "\t 73308055 73308205\n",
- "{'Q.START': 1824, 'Q.END': 1825, 'T.START': 73308053, 'T.END': 73308054, 'CG': '1='}\n",
- "7046637 D134#1#chr03 1975 1976\n",
- "\tNot in path\n",
- "7046638 D134#1#chr03 1976 2015\n",
- "\tIn path\n",
- "\t 73308208 73308247\n",
- "{'Q.START': 1825, 'Q.END': 1975, 'T.START': 73308055, 'T.END': 73308205, 'CG': '150='}\n",
- "7046639 D134#1#chr03 2015 2016\n",
- "\tNot in path\n",
- "7046641 D134#1#chr03 2016 2047\n",
- "\tIn path\n",
- "\t 73308250 73308281\n",
- "{'Q.START': 1976, 'Q.END': 2015, 'T.START': 73308208, 'T.END': 73308247, 'CG': '39='}\n",
- "7046644 D134#1#chr03 2047 2055\n",
- "\tIn path\n",
- "\t 73308286 73308294\n",
- "{'Q.START': 2016, 'Q.END': 2047, 'T.START': 73308250, 'T.END': 73308281, 'CG': '31='}\n",
- "7046646 D134#1#chr03 2055 2056\n",
- "\tNot in path\n",
- "7046647 D134#1#chr03 2056 2120\n",
- "\tIn path\n",
- "\t 73308297 73308361\n",
- "{'Q.START': 2047, 'Q.END': 2055, 'T.START': 73308286, 'T.END': 73308294, 'CG': '8='}\n",
- "7046649 D134#1#chr03 2120 2121\n",
- "\tIn path\n",
- "\t 73308362 73308363\n",
- "{'Q.START': 2056, 'Q.END': 2120, 'T.START': 73308297, 'T.END': 73308361, 'CG': '64='}\n",
- "7046650 D134#1#chr03 2121 2157\n",
- "\tIn path\n",
- "\t 73308364 73308400\n",
- "{'Q.START': 2120, 'Q.END': 2121, 'T.START': 73308362, 'T.END': 73308363, 'CG': '1='}\n",
- "7046652 D134#1#chr03 2157 2158\n",
- "\tNot in path\n",
- "7046653 D134#1#chr03 2158 2170\n",
- "\tIn path\n",
- "\t 73308403 73308415\n",
- "{'Q.START': 2121, 'Q.END': 2157, 'T.START': 73308364, 'T.END': 73308400, 'CG': '36='}\n",
- "7046654 D134#1#chr03 2170 2171\n",
- "\tIn path\n",
- "\t 73308416 73308417\n",
- "{'Q.START': 2158, 'Q.END': 2170, 'T.START': 73308403, 'T.END': 73308415, 'CG': '12='}\n",
- "7046656 D134#1#chr03 2171 2205\n",
- "\tIn path\n",
- "\t 73308418 73308452\n",
- "{'Q.START': 2170, 'Q.END': 2171, 'T.START': 73308416, 'T.END': 73308417, 'CG': '1='}\n",
- "7046657 D134#1#chr03 2205 2206\n",
- "\tNot in path\n",
- "7046659 D134#1#chr03 2206 2344\n",
- "\tIn path\n",
- "\t 73308455 73308593\n",
- "{'Q.START': 2171, 'Q.END': 2205, 'T.START': 73308418, 'T.END': 73308452, 'CG': '34='}\n",
- "7046660 D134#1#chr03 2344 2345\n",
- "\tNot in path\n",
- "7046662 D134#1#chr03 2345 2364\n",
- "\tIn path\n",
- "\t 73308596 73308615\n",
- "{'Q.START': 2206, 'Q.END': 2344, 'T.START': 73308455, 'T.END': 73308593, 'CG': '138='}\n",
- "7046663 D134#1#chr03 2364 2383\n",
- "\tIn path\n",
- "\t 73308616 73308635\n",
- "{'Q.START': 2345, 'Q.END': 2364, 'T.START': 73308596, 'T.END': 73308615, 'CG': '19='}\n",
- "7046665 D134#1#chr03 2383 2408\n",
- "\tIn path\n",
- "\t 73308636 73308661\n",
- "{'Q.START': 2364, 'Q.END': 2383, 'T.START': 73308616, 'T.END': 73308635, 'CG': '19='}\n",
- "7046667 D134#1#chr03 2408 2409\n",
- "\tIn path\n",
- "\t 73308662 73308663\n",
- "{'Q.START': 2383, 'Q.END': 2408, 'T.START': 73308636, 'T.END': 73308661, 'CG': '25='}\n",
- "7046668 D134#1#chr03 2409 2441\n",
- "\tIn path\n",
- "\t 73308664 73308696\n",
- "{'Q.START': 2408, 'Q.END': 2409, 'T.START': 73308662, 'T.END': 73308663, 'CG': '1='}\n",
- "7046670 D134#1#chr03 2441 2442\n",
- "\tIn path\n",
- "\t 73308697 73308698\n",
- "{'Q.START': 2409, 'Q.END': 2441, 'T.START': 73308664, 'T.END': 73308696, 'CG': '32='}\n",
- "7046671 D134#1#chr03 2442 2580\n",
- "\tIn path\n",
- "\t 73308699 73308837\n",
- "{'Q.START': 2441, 'Q.END': 2442, 'T.START': 73308697, 'T.END': 73308698, 'CG': '1='}\n",
- "7046674 D134#1#chr03 2582 2583\n",
- "\tIn path\n",
- "\t 73308838 73308839\n",
- "{'Q.START': 2442, 'Q.END': 2580, 'T.START': 73308699, 'T.END': 73308837, 'CG': '138='}\n",
- "7046675 D134#1#chr03 2583 2584\n",
- "\tIn path\n",
- "\t 73308840 73308841\n",
- "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}\n",
- "7046674 D134#1#chr03 2582 2583\n",
- "\tIn path\n",
- "\t 73308838 73308839\n",
- "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}\n",
- "7046675 D134#1#chr03 2583 2584\n",
- "\tIn path\n",
- "\t 73308840 73308841\n",
- "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}\n",
- "7046676 D134#1#chr03 2584 2764\n",
- "\tIn path\n",
- "\t 73308842 73309022\n",
- "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}\n",
- "7046678 D134#1#chr03 2764 2765\n",
- "\tNot in path\n",
- "7046679 D134#1#chr03 2765 2797\n",
- "\tIn path\n",
- "\t 73309025 73309057\n",
- "{'Q.START': 2584, 'Q.END': 2764, 'T.START': 73308842, 'T.END': 73309022, 'CG': '180='}\n",
- "7046680 D134#1#chr03 2797 2798\n",
- "\tNot in path\n",
- "7046682 D134#1#chr03 2798 2878\n",
- "\tIn path\n",
- "\t 73309060 73309140\n",
- "{'Q.START': 2765, 'Q.END': 2797, 'T.START': 73309025, 'T.END': 73309057, 'CG': '32='}\n",
- "7046684 D134#1#chr03 2878 2879\n",
- "\tIn path\n",
- "\t 73309141 73309142\n",
- "{'Q.START': 2798, 'Q.END': 2878, 'T.START': 73309060, 'T.END': 73309140, 'CG': '80='}\n",
- "7046685 D134#1#chr03 2879 2951\n",
- "\tIn path\n",
- "\t 73309143 73309215\n",
- "{'Q.START': 2878, 'Q.END': 2879, 'T.START': 73309141, 'T.END': 73309142, 'CG': '1='}\n",
- "7046686 D134#1#chr03 2951 2952\n",
- "\tIn path\n",
- "\t 73309216 73309217\n",
- "{'Q.START': 2879, 'Q.END': 2951, 'T.START': 73309143, 'T.END': 73309215, 'CG': '72='}\n",
- "7046688 D134#1#chr03 2952 3002\n",
- "\tIn path\n",
- "\t 73309218 73309268\n",
- "{'Q.START': 2951, 'Q.END': 2952, 'T.START': 73309216, 'T.END': 73309217, 'CG': '1='}\n",
- "7046690 D134#1#chr03 3002 3077\n",
- "\tIn path\n",
- "\t 73309271 73309346\n",
- "{'Q.START': 2952, 'Q.END': 3002, 'T.START': 73309218, 'T.END': 73309268, 'CG': '50='}\n",
- "7046692 D134#1#chr03 3077 3078\n",
- "\tIn path\n",
- "\t 73309347 73309348\n",
- "{'Q.START': 3002, 'Q.END': 3077, 'T.START': 73309271, 'T.END': 73309346, 'CG': '75='}\n",
- "7046693 D134#1#chr03 3078 3093\n",
- "\tIn path\n",
- "\t 73309349 73309364\n",
- "{'Q.START': 3077, 'Q.END': 3078, 'T.START': 73309347, 'T.END': 73309348, 'CG': '1='}\n",
- "7046695 D134#1#chr03 3093 3094\n",
- "\tNot in path\n",
- "7046696 D134#1#chr03 3094 3097\n",
- "\tIn path\n",
- "\t 73309367 73309370\n",
- "{'Q.START': 3078, 'Q.END': 3093, 'T.START': 73309349, 'T.END': 73309364, 'CG': '15='}\n",
- "7046698 D134#1#chr03 3097 3140\n",
- "\tIn path\n",
- "\t 73309371 73309414\n",
- "{'Q.START': 3094, 'Q.END': 3097, 'T.START': 73309367, 'T.END': 73309370, 'CG': '3='}\n",
- "7046700 D134#1#chr03 3140 3210\n",
- "\tIn path\n",
- "\t 73309415 73309485\n",
- "{'Q.START': 3097, 'Q.END': 3140, 'T.START': 73309371, 'T.END': 73309414, 'CG': '43='}\n",
- "7046702 D134#1#chr03 3210 3211\n",
- "\tIn path\n",
- "\t 73309486 73309487\n",
- "{'Q.START': 3140, 'Q.END': 3210, 'T.START': 73309415, 'T.END': 73309485, 'CG': '70='}\n",
- "7046703 D134#1#chr03 3211 3229\n",
- "\tIn path\n",
- "\t 73309488 73309506\n",
- "{'Q.START': 3210, 'Q.END': 3211, 'T.START': 73309486, 'T.END': 73309487, 'CG': '1='}\n",
- "7046704 D134#1#chr03 3229 3230\n",
- "\tIn path\n",
- "\t 73309507 73309508\n",
- "{'Q.START': 3211, 'Q.END': 3229, 'T.START': 73309488, 'T.END': 73309506, 'CG': '18='}\n",
- "7046706 D134#1#chr03 3230 3276\n",
- "\tIn path\n",
- "\t 73309509 73309555\n",
- "{'Q.START': 3229, 'Q.END': 3230, 'T.START': 73309507, 'T.END': 73309508, 'CG': '1='}\n",
- "7046707 D134#1#chr03 3276 3277\n",
- "\tNot in path\n",
- "7046709 D134#1#chr03 3277 3315\n",
- "\tIn path\n",
- "\t 73309558 73309596\n",
- "{'Q.START': 3230, 'Q.END': 3276, 'T.START': 73309509, 'T.END': 73309555, 'CG': '46='}\n",
- "7046710 D134#1#chr03 3315 3316\n",
- "\tNot in path\n",
- "7046712 D134#1#chr03 3316 3322\n",
- "\tIn path\n",
- "\t 73309599 73309605\n",
- "{'Q.START': 3277, 'Q.END': 3315, 'T.START': 73309558, 'T.END': 73309596, 'CG': '38='}\n",
- "7046713 D134#1#chr03 3322 3323\n",
- "\tNot in path\n",
- "7046715 D134#1#chr03 3323 3348\n",
- "\tIn path\n",
- "\t 73309608 73309633\n",
- "{'Q.START': 3316, 'Q.END': 3322, 'T.START': 73309599, 'T.END': 73309605, 'CG': '6='}\n",
- "7046718 D134#1#chr03 3352 3353\n",
- "\tIn path\n",
- "\t 73309634 73309635\n",
- "{'Q.START': 3323, 'Q.END': 3348, 'T.START': 73309608, 'T.END': 73309633, 'CG': '25='}\n",
- "7046717 D134#1#chr03 3351 3352\n",
- "\tIn path\n",
- "\t 73309636 73309637\n",
- "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
- "7046718 D134#1#chr03 3352 3353\n",
- "\tIn path\n",
- "\t 73309634 73309635\n",
- "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}\n",
- "7046717 D134#1#chr03 3351 3352\n",
- "\tIn path\n",
- "\t 73309636 73309637\n",
- "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
- "7046718 D134#1#chr03 3352 3353\n",
- "\tIn path\n",
- "\t 73309634 73309635\n",
- "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}\n",
- "7046720 D134#1#chr03 3353 3354\n",
- "\tIn path\n",
- "\t 73309638 73309639\n",
- "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}\n",
- "7046722 D134#1#chr03 3354 3356\n",
- "\tIn path\n",
- "\t 73309640 73309642\n",
- "{'Q.START': 3353, 'Q.END': 3354, 'T.START': 73309638, 'T.END': 73309639, 'CG': '1='}\n",
- "7046724 D134#1#chr03 3356 3357\n",
- "\tNot in path\n",
- "7046725 D134#1#chr03 3357 3489\n",
- "\tIn path\n",
- "\t 73309645 73309777\n",
- "{'Q.START': 3354, 'Q.END': 3356, 'T.START': 73309640, 'T.END': 73309642, 'CG': '2='}\n",
- "7046727 D134#1#chr03 3489 3490\n",
- "\tNot in path\n",
- "7046728 D134#1#chr03 3490 3642\n",
- "\tIn path\n",
- "\t 73309780 73309932\n",
- "{'Q.START': 3357, 'Q.END': 3489, 'T.START': 73309645, 'T.END': 73309777, 'CG': '132='}\n",
- "7046729 D134#1#chr03 3642 3644\n",
- "\tNot in path\n",
- "7046730 D134#1#chr03 3644 3685\n",
- "\tIn path\n",
- "\t 73309933 73309974\n",
- "{'Q.START': 3490, 'Q.END': 3642, 'T.START': 73309780, 'T.END': 73309932, 'CG': '152='}\n",
- "7046731 D134#1#chr03 3685 3687\n",
- "\tNot in path\n",
- "7046733 D134#1#chr03 3687 3693\n",
- "\tIn path\n",
- "\t 73309977 73309983\n",
- "{'Q.START': 3644, 'Q.END': 3685, 'T.START': 73309933, 'T.END': 73309974, 'CG': '41='}\n",
- "7046735 D134#1#chr03 3693 3694\n",
- "\tNot in path\n",
- "7046736 D134#1#chr03 3694 3708\n",
- "\tIn path\n",
- "\t 73309986 73310000\n",
- "{'Q.START': 3687, 'Q.END': 3693, 'T.START': 73309977, 'T.END': 73309983, 'CG': '6='}\n",
- "7046738 D134#1#chr03 3720 3721\n",
- "\tIn path\n",
- "\t 73310010 73310011\n",
- "{'Q.START': 3694, 'Q.END': 3708, 'T.START': 73309986, 'T.END': 73310000, 'CG': '14='}\n",
- "7046739 D134#1#chr03 3721 3722\n",
- "\tIn path\n",
- "\t 73310003 73310004\n",
- "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
- "7046740 D134#1#chr03 3716 3720\n",
- "\tIn path\n",
- "\t 73310005 73310009\n",
- "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
- "7046738 D134#1#chr03 3720 3721\n",
- "\tIn path\n",
- "\t 73310010 73310011\n",
- "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}\n",
- "7046739 D134#1#chr03 3721 3722\n",
- "\tIn path\n",
- "\t 73310003 73310004\n",
- "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
- "7046740 D134#1#chr03 3716 3720\n",
- "\tIn path\n",
- "\t 73310005 73310009\n",
- "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
- "7046738 D134#1#chr03 3720 3721\n",
- "\tIn path\n",
- "\t 73310010 73310011\n",
- "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}\n",
- "7046739 D134#1#chr03 3721 3722\n",
- "\tIn path\n",
- "\t 73310003 73310004\n",
- "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}\n",
- "7046741 D134#1#chr03 3722 3735\n",
- "\tIn path\n",
- "\t 73310012 73310045\n",
- "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}\n",
- "ALN_1\n",
- "7046526 TO1000#1#chr03 0 77\n",
- "\t 64684013 64684090\n",
- "skipped\n",
- "\n",
- "7046528 TO1000#1#chr03 77 82\n",
- "\t 64684091 64684096\n",
- "{'Q.START': 0, 'Q.END': 77, 'T.START': 64684013, 'T.END': 64684090, 'CG': '77='}\n",
- "7046530 TO1000#1#chr03 82 83\n",
- "\t 64684097 64684098\n",
- "{'Q.START': 77, 'Q.END': 82, 'T.START': 64684091, 'T.END': 64684096, 'CG': '5='}\n",
- "7046531 TO1000#1#chr03 83 138\n",
- "\t 64684099 64684154\n",
- "{'Q.START': 82, 'Q.END': 83, 'T.START': 64684097, 'T.END': 64684098, 'CG': '1='}\n",
- "7046532 TO1000#1#chr03 138 139\n",
- "\t 64684155 64684156\n",
- "{'Q.START': 83, 'Q.END': 138, 'T.START': 64684099, 'T.END': 64684154, 'CG': '55='}\n",
- "7046533 TO1000#1#chr03 139 202\n",
- "\t 64684157 64684220\n",
- "{'Q.START': 138, 'Q.END': 139, 'T.START': 64684155, 'T.END': 64684156, 'CG': '1='}\n",
- "7046534 TO1000#1#chr03 202 203\n",
- "\t 64684221 64684222\n",
- "{'Q.START': 139, 'Q.END': 202, 'T.START': 64684157, 'T.END': 64684220, 'CG': '63='}\n",
- "7046536 TO1000#1#chr03 203 379\n",
- "\t 64684223 64684399\n",
- "{'Q.START': 202, 'Q.END': 203, 'T.START': 64684221, 'T.END': 64684222, 'CG': '1='}\n",
- "7046537 TO1000#1#chr03 379 380\n",
- "\t 64684400 64684401\n",
- "{'Q.START': 203, 'Q.END': 379, 'T.START': 64684223, 'T.END': 64684399, 'CG': '176='}\n",
- "7046539 TO1000#1#chr03 380 429\n",
- "\t 64684402 64684451\n",
- "{'Q.START': 379, 'Q.END': 380, 'T.START': 64684400, 'T.END': 64684401, 'CG': '1='}\n",
- "7046541 TO1000#1#chr03 429 430\n",
- "\t 64684452 64684453\n",
- "{'Q.START': 380, 'Q.END': 429, 'T.START': 64684402, 'T.END': 64684451, 'CG': '49='}\n",
- "7046542 TO1000#1#chr03 430 457\n",
- "\t 64684454 64684481\n",
- "{'Q.START': 429, 'Q.END': 430, 'T.START': 64684452, 'T.END': 64684453, 'CG': '1='}\n",
- "7046544 TO1000#1#chr03 457 492\n",
- "\t 64684482 64684517\n",
- "{'Q.START': 430, 'Q.END': 457, 'T.START': 64684454, 'T.END': 64684481, 'CG': '27='}\n",
- "7046546 TO1000#1#chr03 492 494\n",
- "\t 64684518 64684520\n",
- "{'Q.START': 457, 'Q.END': 492, 'T.START': 64684482, 'T.END': 64684517, 'CG': '35='}\n",
- "7046547 TO1000#1#chr03 494 497\n",
- "\t 64684521 64684524\n",
- "{'Q.START': 492, 'Q.END': 494, 'T.START': 64684518, 'T.END': 64684520, 'CG': '2='}\n",
- "7046549 TO1000#1#chr03 497 507\n",
- "\t 64684525 64684535\n",
- "{'Q.START': 494, 'Q.END': 497, 'T.START': 64684521, 'T.END': 64684524, 'CG': '3='}\n",
- "7046551 TO1000#1#chr03 507 508\n",
- "\t 64684536 64684537\n",
- "{'Q.START': 497, 'Q.END': 507, 'T.START': 64684525, 'T.END': 64684535, 'CG': '10='}\n",
- "7046552 TO1000#1#chr03 508 564\n",
- "\t 64684538 64684594\n",
- "{'Q.START': 507, 'Q.END': 508, 'T.START': 64684536, 'T.END': 64684537, 'CG': '1='}\n",
- "7046554 TO1000#1#chr03 564 566\n",
- "\t 64684595 64684597\n",
- "{'Q.START': 508, 'Q.END': 564, 'T.START': 64684538, 'T.END': 64684594, 'CG': '56='}\n",
- "7046556 TO1000#1#chr03 568 569\n",
- "\t 64684598 64684599\n",
- "{'Q.START': 564, 'Q.END': 566, 'T.START': 64684595, 'T.END': 64684597, 'CG': '2='}\n",
- "7046556 TO1000#1#chr03 568 569\n",
- "\t 64684598 64684599\n",
- "{'Q.START': 568, 'Q.END': 569, 'T.START': 64684598, 'T.END': 64684599, 'CG': '1='}\n",
- "7046556 TO1000#1#chr03 568 569\n",
- "\t 64684598 64684599\n",
- "{'Q.START': 568, 'Q.END': 569, 'T.START': 64684598, 'T.END': 64684599, 'CG': '1='}\n",
- "7046557 TO1000#1#chr03 569 824\n",
- "\t 64684600 64684855\n",
- "{'Q.START': 568, 'Q.END': 569, 'T.START': 64684598, 'T.END': 64684599, 'CG': '1='}\n",
- "7046558 TO1000#1#chr03 824 826\n",
- "\t 64684856 64684858\n",
- "{'Q.START': 569, 'Q.END': 824, 'T.START': 64684600, 'T.END': 64684855, 'CG': '255='}\n",
- "7046559 TO1000#1#chr03 826 858\n",
- "\t 64684859 64684891\n",
- "{'Q.START': 824, 'Q.END': 826, 'T.START': 64684856, 'T.END': 64684858, 'CG': '2='}\n",
- "7046560 TO1000#1#chr03 858 859\n",
- "\t 64684892 64684893\n",
- "{'Q.START': 826, 'Q.END': 858, 'T.START': 64684859, 'T.END': 64684891, 'CG': '32='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 858, 'Q.END': 859, 'T.START': 64684892, 'T.END': 64684893, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046561 TO1000#1#chr03 868 869\n",
- "\t 64684894 64684895\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046562 TO1000#1#chr03 869 913\n",
- "\t 64684896 64684940\n",
- "{'Q.START': 868, 'Q.END': 869, 'T.START': 64684894, 'T.END': 64684895, 'CG': '1='}\n",
- "7046564 TO1000#1#chr03 913 919\n",
- "\t 64684941 64684947\n",
- "{'Q.START': 869, 'Q.END': 913, 'T.START': 64684896, 'T.END': 64684940, 'CG': '44='}\n",
- "7046565 TO1000#1#chr03 919 978\n",
- "\t 64684948 64685007\n",
- "{'Q.START': 913, 'Q.END': 919, 'T.START': 64684941, 'T.END': 64684947, 'CG': '6='}\n",
- "7046567 TO1000#1#chr03 978 979\n",
- "\t 64685008 64685009\n",
- "{'Q.START': 919, 'Q.END': 978, 'T.START': 64684948, 'T.END': 64685007, 'CG': '59='}\n",
- "7046568 TO1000#1#chr03 979 1038\n",
- "\t 64685010 64685069\n",
- "{'Q.START': 978, 'Q.END': 979, 'T.START': 64685008, 'T.END': 64685009, 'CG': '1='}\n",
- "7046570 TO1000#1#chr03 1038 1045\n",
- "\t 64685070 64685077\n",
- "{'Q.START': 979, 'Q.END': 1038, 'T.START': 64685010, 'T.END': 64685069, 'CG': '59='}\n",
- "7046571 TO1000#1#chr03 1045 1046\n",
- "\t 64685078 64685079\n",
- "{'Q.START': 1038, 'Q.END': 1045, 'T.START': 64685070, 'T.END': 64685077, 'CG': '7='}\n",
- "7046573 TO1000#1#chr03 1046 1080\n",
- "\t 64685080 64685114\n",
- "{'Q.START': 1045, 'Q.END': 1046, 'T.START': 64685078, 'T.END': 64685079, 'CG': '1='}\n",
- "7046574 TO1000#1#chr03 1080 1081\n",
- "\t 64685115 64685116\n",
- "{'Q.START': 1046, 'Q.END': 1080, 'T.START': 64685080, 'T.END': 64685114, 'CG': '34='}\n",
- "7046576 TO1000#1#chr03 1081 1107\n",
- "\t 64685117 64685143\n",
- "{'Q.START': 1080, 'Q.END': 1081, 'T.START': 64685115, 'T.END': 64685116, 'CG': '1='}\n",
- "7046577 TO1000#1#chr03 1107 1108\n",
- "\t 64685144 64685145\n",
- "{'Q.START': 1081, 'Q.END': 1107, 'T.START': 64685117, 'T.END': 64685143, 'CG': '26='}\n",
- "7046579 TO1000#1#chr03 1108 1183\n",
- "\t 64685146 64685221\n",
- "{'Q.START': 1107, 'Q.END': 1108, 'T.START': 64685144, 'T.END': 64685145, 'CG': '1='}\n",
- "7046581 TO1000#1#chr03 1183 1186\n",
- "\t 64685222 64685225\n",
- "{'Q.START': 1108, 'Q.END': 1183, 'T.START': 64685146, 'T.END': 64685221, 'CG': '75='}\n",
- "7046583 TO1000#1#chr03 1186 1224\n",
- "\t 64685226 64685264\n",
- "{'Q.START': 1183, 'Q.END': 1186, 'T.START': 64685222, 'T.END': 64685225, 'CG': '3='}\n",
- "7046584 TO1000#1#chr03 1224 1257\n",
- "\t 64685265 64685298\n",
- "{'Q.START': 1186, 'Q.END': 1224, 'T.START': 64685226, 'T.END': 64685264, 'CG': '38='}\n",
- "7046586 TO1000#1#chr03 1257 1289\n",
- "\t 64685299 64685331\n",
- "{'Q.START': 1224, 'Q.END': 1257, 'T.START': 64685265, 'T.END': 64685298, 'CG': '33='}\n",
- "7046587 TO1000#1#chr03 1289 1311\n",
- "\t 64685332 64685354\n",
- "{'Q.START': 1257, 'Q.END': 1289, 'T.START': 64685299, 'T.END': 64685331, 'CG': '32='}\n",
- "7046589 TO1000#1#chr03 1311 1359\n",
- "\t 64685355 64685403\n",
- "{'Q.START': 1289, 'Q.END': 1311, 'T.START': 64685332, 'T.END': 64685354, 'CG': '22='}\n",
- "7046590 TO1000#1#chr03 1359 1382\n",
- "\t 64685404 64685427\n",
- "{'Q.START': 1311, 'Q.END': 1359, 'T.START': 64685355, 'T.END': 64685403, 'CG': '48='}\n",
- "7046592 TO1000#1#chr03 1382 1434\n",
- "\t 64685428 64685480\n",
- "{'Q.START': 1359, 'Q.END': 1382, 'T.START': 64685404, 'T.END': 64685427, 'CG': '23='}\n",
- "7046593 TO1000#1#chr03 1434 1451\n",
- "\t 64685481 64685498\n",
- "{'Q.START': 1382, 'Q.END': 1434, 'T.START': 64685428, 'T.END': 64685480, 'CG': '52='}\n",
- "7046594 TO1000#1#chr03 1451 1531\n",
- "\t 64685499 64685579\n",
- "{'Q.START': 1434, 'Q.END': 1451, 'T.START': 64685481, 'T.END': 64685498, 'CG': '17='}\n",
- "7046596 TO1000#1#chr03 1531 1532\n",
- "\t 64685580 64685581\n",
- "{'Q.START': 1451, 'Q.END': 1531, 'T.START': 64685499, 'T.END': 64685579, 'CG': '80='}\n",
- "7046597 TO1000#1#chr03 1532 1543\n",
- "\t 64685582 64685593\n",
- "{'Q.START': 1531, 'Q.END': 1532, 'T.START': 64685580, 'T.END': 64685581, 'CG': '1='}\n",
- "7046599 TO1000#1#chr03 1543 1544\n",
- "\t 64685594 64685595\n",
- "{'Q.START': 1532, 'Q.END': 1543, 'T.START': 64685582, 'T.END': 64685593, 'CG': '11='}\n",
- "7046600 TO1000#1#chr03 1544 1572\n",
- "\t 64685596 64685624\n",
- "{'Q.START': 1543, 'Q.END': 1544, 'T.START': 64685594, 'T.END': 64685595, 'CG': '1='}\n",
- "7046601 TO1000#1#chr03 1572 1573\n",
- "\t 64685625 64685626\n",
- "{'Q.START': 1544, 'Q.END': 1572, 'T.START': 64685596, 'T.END': 64685624, 'CG': '28='}\n",
- "7046603 TO1000#1#chr03 1573 1587\n",
- "\t 64685627 64685641\n",
- "{'Q.START': 1572, 'Q.END': 1573, 'T.START': 64685625, 'T.END': 64685626, 'CG': '1='}\n",
- "7046604 TO1000#1#chr03 1587 1588\n",
- "\t 64685642 64685643\n",
- "{'Q.START': 1573, 'Q.END': 1587, 'T.START': 64685627, 'T.END': 64685641, 'CG': '14='}\n",
- "7046606 TO1000#1#chr03 1588 1616\n",
- "\t 64685644 64685672\n",
- "{'Q.START': 1587, 'Q.END': 1588, 'T.START': 64685642, 'T.END': 64685643, 'CG': '1='}\n",
- "7046608 TO1000#1#chr03 1616 1617\n",
- "\t 64685673 64685674\n",
- "{'Q.START': 1588, 'Q.END': 1616, 'T.START': 64685644, 'T.END': 64685672, 'CG': '28='}\n",
- "7046609 TO1000#1#chr03 1617 1646\n",
- "\t 64685675 64685704\n",
- "{'Q.START': 1616, 'Q.END': 1617, 'T.START': 64685673, 'T.END': 64685674, 'CG': '1='}\n",
- "7046621 TO1000#1#chr03 1646 1661\n",
- "\t 64685705 64685720\n",
- "{'Q.START': 1617, 'Q.END': 1646, 'T.START': 64685675, 'T.END': 64685704, 'CG': '29='}\n",
- "7046622 TO1000#1#chr03 1661 1673\n",
- "\t 64685721 64685733\n",
- "{'Q.START': 1646, 'Q.END': 1661, 'T.START': 64685705, 'T.END': 64685720, 'CG': '15='}\n",
- "7046624 TO1000#1#chr03 1673 1674\n",
- "\t 64685734 64685735\n",
- "{'Q.START': 1661, 'Q.END': 1673, 'T.START': 64685721, 'T.END': 64685733, 'CG': '12='}\n",
- "7046625 TO1000#1#chr03 1674 1726\n",
- "\t 64685736 64685788\n",
- "{'Q.START': 1673, 'Q.END': 1674, 'T.START': 64685734, 'T.END': 64685735, 'CG': '1='}\n",
- "7046626 TO1000#1#chr03 1726 1727\n",
- "\t 64685789 64685790\n",
- "{'Q.START': 1674, 'Q.END': 1726, 'T.START': 64685736, 'T.END': 64685788, 'CG': '52='}\n",
- "7046628 TO1000#1#chr03 1727 1762\n",
- "\t 64685791 64685826\n",
- "{'Q.START': 1726, 'Q.END': 1727, 'T.START': 64685789, 'T.END': 64685790, 'CG': '1='}\n",
- "7046631 TO1000#1#chr03 1766 1767\n",
- "\t 64685827 64685828\n",
- "{'Q.START': 1727, 'Q.END': 1762, 'T.START': 64685791, 'T.END': 64685826, 'CG': '35='}\n",
- "7046673 TO1000#1#chr03 1765 1766\n",
- "\t 64685829 64685830\n",
- "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 64685827, 'T.END': 64685828, 'CG': '1='}\n",
- "7046631 TO1000#1#chr03 1766 1767\n",
- "\t 64685827 64685828\n",
- "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 64685829, 'T.END': 64685830, 'CG': '1='}\n",
- "7046673 TO1000#1#chr03 1765 1766\n",
- "\t 64685829 64685830\n",
- "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 64685827, 'T.END': 64685828, 'CG': '1='}\n",
- "7046631 TO1000#1#chr03 1766 1767\n",
- "\t 64685827 64685828\n",
- "{'Q.START': 1765, 'Q.END': 1766, 'T.START': 64685829, 'T.END': 64685830, 'CG': '1='}\n",
- "7046632 TO1000#1#chr03 1767 1824\n",
- "\t 64685831 64685888\n",
- "{'Q.START': 1766, 'Q.END': 1767, 'T.START': 64685827, 'T.END': 64685828, 'CG': '1='}\n",
- "7046634 TO1000#1#chr03 1824 1825\n",
- "\t 64685889 64685890\n",
- "{'Q.START': 1767, 'Q.END': 1824, 'T.START': 64685831, 'T.END': 64685888, 'CG': '57='}\n",
- "7046635 TO1000#1#chr03 1825 1975\n",
- "\t 64685891 64686041\n",
- "{'Q.START': 1824, 'Q.END': 1825, 'T.START': 64685889, 'T.END': 64685890, 'CG': '1='}\n",
- "7046637 TO1000#1#chr03 1975 1976\n",
- "\t 64686042 64686043\n",
- "{'Q.START': 1825, 'Q.END': 1975, 'T.START': 64685891, 'T.END': 64686041, 'CG': '150='}\n",
- "7046638 TO1000#1#chr03 1976 2015\n",
- "\t 64686044 64686083\n",
- "{'Q.START': 1975, 'Q.END': 1976, 'T.START': 64686042, 'T.END': 64686043, 'CG': '1='}\n",
- "7046639 TO1000#1#chr03 2015 2016\n",
- "\t 64686084 64686085\n",
- "{'Q.START': 1976, 'Q.END': 2015, 'T.START': 64686044, 'T.END': 64686083, 'CG': '39='}\n",
- "7046641 TO1000#1#chr03 2016 2047\n",
- "\t 64686086 64686117\n",
- "{'Q.START': 2015, 'Q.END': 2016, 'T.START': 64686084, 'T.END': 64686085, 'CG': '1='}\n",
- "7046644 TO1000#1#chr03 2047 2055\n",
- "\t 64686118 64686126\n",
- "{'Q.START': 2016, 'Q.END': 2047, 'T.START': 64686086, 'T.END': 64686117, 'CG': '31='}\n",
- "7046646 TO1000#1#chr03 2055 2056\n",
- "\t 64686127 64686128\n",
- "{'Q.START': 2047, 'Q.END': 2055, 'T.START': 64686118, 'T.END': 64686126, 'CG': '8='}\n",
- "7046647 TO1000#1#chr03 2056 2120\n",
- "\t 64686129 64686193\n",
- "{'Q.START': 2055, 'Q.END': 2056, 'T.START': 64686127, 'T.END': 64686128, 'CG': '1='}\n",
- "7046649 TO1000#1#chr03 2120 2121\n",
- "\t 64686194 64686195\n",
- "{'Q.START': 2056, 'Q.END': 2120, 'T.START': 64686129, 'T.END': 64686193, 'CG': '64='}\n",
- "7046650 TO1000#1#chr03 2121 2157\n",
- "\t 64686196 64686232\n",
- "{'Q.START': 2120, 'Q.END': 2121, 'T.START': 64686194, 'T.END': 64686195, 'CG': '1='}\n",
- "7046652 TO1000#1#chr03 2157 2158\n",
- "\t 64686233 64686234\n",
- "{'Q.START': 2121, 'Q.END': 2157, 'T.START': 64686196, 'T.END': 64686232, 'CG': '36='}\n",
- "7046653 TO1000#1#chr03 2158 2170\n",
- "\t 64686235 64686247\n",
- "{'Q.START': 2157, 'Q.END': 2158, 'T.START': 64686233, 'T.END': 64686234, 'CG': '1='}\n",
- "7046654 TO1000#1#chr03 2170 2171\n",
- "\t 64686248 64686249\n",
- "{'Q.START': 2158, 'Q.END': 2170, 'T.START': 64686235, 'T.END': 64686247, 'CG': '12='}\n",
- "7046656 TO1000#1#chr03 2171 2205\n",
- "\t 64686250 64686284\n",
- "{'Q.START': 2170, 'Q.END': 2171, 'T.START': 64686248, 'T.END': 64686249, 'CG': '1='}\n",
- "7046657 TO1000#1#chr03 2205 2206\n",
- "\t 64686285 64686286\n",
- "{'Q.START': 2171, 'Q.END': 2205, 'T.START': 64686250, 'T.END': 64686284, 'CG': '34='}\n",
- "7046659 TO1000#1#chr03 2206 2344\n",
- "\t 64686287 64686425\n",
- "{'Q.START': 2205, 'Q.END': 2206, 'T.START': 64686285, 'T.END': 64686286, 'CG': '1='}\n",
- "7046660 TO1000#1#chr03 2344 2345\n",
- "\t 64686426 64686427\n",
- "{'Q.START': 2206, 'Q.END': 2344, 'T.START': 64686287, 'T.END': 64686425, 'CG': '138='}\n",
- "7046662 TO1000#1#chr03 2345 2364\n",
- "\t 64686428 64686447\n",
- "{'Q.START': 2344, 'Q.END': 2345, 'T.START': 64686426, 'T.END': 64686427, 'CG': '1='}\n",
- "7046663 TO1000#1#chr03 2364 2383\n",
- "\t 64686448 64686467\n",
- "{'Q.START': 2345, 'Q.END': 2364, 'T.START': 64686428, 'T.END': 64686447, 'CG': '19='}\n",
- "7046665 TO1000#1#chr03 2383 2408\n",
- "\t 64686468 64686493\n",
- "{'Q.START': 2364, 'Q.END': 2383, 'T.START': 64686448, 'T.END': 64686467, 'CG': '19='}\n",
- "7046667 TO1000#1#chr03 2408 2409\n",
- "\t 64686494 64686495\n",
- "{'Q.START': 2383, 'Q.END': 2408, 'T.START': 64686468, 'T.END': 64686493, 'CG': '25='}\n",
- "7046668 TO1000#1#chr03 2409 2441\n",
- "\t 64686496 64686528\n",
- "{'Q.START': 2408, 'Q.END': 2409, 'T.START': 64686494, 'T.END': 64686495, 'CG': '1='}\n",
- "7046670 TO1000#1#chr03 2441 2442\n",
- "\t 64686529 64686530\n",
- "{'Q.START': 2409, 'Q.END': 2441, 'T.START': 64686496, 'T.END': 64686528, 'CG': '32='}\n",
- "7046671 TO1000#1#chr03 2442 2580\n",
- "\t 64686531 64686669\n",
- "{'Q.START': 2441, 'Q.END': 2442, 'T.START': 64686529, 'T.END': 64686530, 'CG': '1='}\n",
- "7046674 TO1000#1#chr03 2582 2583\n",
- "\t 64686670 64686671\n",
- "{'Q.START': 2442, 'Q.END': 2580, 'T.START': 64686531, 'T.END': 64686669, 'CG': '138='}\n",
- "7046675 TO1000#1#chr03 2583 2584\n",
- "\t 64686672 64686673\n",
- "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 64686670, 'T.END': 64686671, 'CG': '1='}\n",
- "7046674 TO1000#1#chr03 2582 2583\n",
- "\t 64686670 64686671\n",
- "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 64686672, 'T.END': 64686673, 'CG': '1='}\n",
- "7046675 TO1000#1#chr03 2583 2584\n",
- "\t 64686672 64686673\n",
- "{'Q.START': 2582, 'Q.END': 2583, 'T.START': 64686670, 'T.END': 64686671, 'CG': '1='}\n",
- "7046676 TO1000#1#chr03 2584 2764\n",
- "\t 64686674 64686854\n",
- "{'Q.START': 2583, 'Q.END': 2584, 'T.START': 64686672, 'T.END': 64686673, 'CG': '1='}\n",
- "7046678 TO1000#1#chr03 2764 2765\n",
- "\t 64686855 64686856\n",
- "{'Q.START': 2584, 'Q.END': 2764, 'T.START': 64686674, 'T.END': 64686854, 'CG': '180='}\n",
- "7046679 TO1000#1#chr03 2765 2797\n",
- "\t 64686857 64686889\n",
- "{'Q.START': 2764, 'Q.END': 2765, 'T.START': 64686855, 'T.END': 64686856, 'CG': '1='}\n",
- "7046680 TO1000#1#chr03 2797 2798\n",
- "\t 64686890 64686891\n",
- "{'Q.START': 2765, 'Q.END': 2797, 'T.START': 64686857, 'T.END': 64686889, 'CG': '32='}\n",
- "7046682 TO1000#1#chr03 2798 2878\n",
- "\t 64686892 64686972\n",
- "{'Q.START': 2797, 'Q.END': 2798, 'T.START': 64686890, 'T.END': 64686891, 'CG': '1='}\n",
- "7046684 TO1000#1#chr03 2878 2879\n",
- "\t 64686973 64686974\n",
- "{'Q.START': 2798, 'Q.END': 2878, 'T.START': 64686892, 'T.END': 64686972, 'CG': '80='}\n",
- "7046685 TO1000#1#chr03 2879 2951\n",
- "\t 64686975 64687047\n",
- "{'Q.START': 2878, 'Q.END': 2879, 'T.START': 64686973, 'T.END': 64686974, 'CG': '1='}\n",
- "7046686 TO1000#1#chr03 2951 2952\n",
- "\t 64687048 64687049\n",
- "{'Q.START': 2879, 'Q.END': 2951, 'T.START': 64686975, 'T.END': 64687047, 'CG': '72='}\n",
- "7046688 TO1000#1#chr03 2952 3002\n",
- "\t 64687050 64687100\n",
- "{'Q.START': 2951, 'Q.END': 2952, 'T.START': 64687048, 'T.END': 64687049, 'CG': '1='}\n",
- "7046690 TO1000#1#chr03 3002 3077\n",
- "\t 64687101 64687176\n",
- "{'Q.START': 2952, 'Q.END': 3002, 'T.START': 64687050, 'T.END': 64687100, 'CG': '50='}\n",
- "7046692 TO1000#1#chr03 3077 3078\n",
- "\t 64687177 64687178\n",
- "{'Q.START': 3002, 'Q.END': 3077, 'T.START': 64687101, 'T.END': 64687176, 'CG': '75='}\n",
- "7046693 TO1000#1#chr03 3078 3093\n",
- "\t 64687179 64687194\n",
- "{'Q.START': 3077, 'Q.END': 3078, 'T.START': 64687177, 'T.END': 64687178, 'CG': '1='}\n",
- "7046695 TO1000#1#chr03 3093 3094\n",
- "\t 64687195 64687196\n",
- "{'Q.START': 3078, 'Q.END': 3093, 'T.START': 64687179, 'T.END': 64687194, 'CG': '15='}\n",
- "7046696 TO1000#1#chr03 3094 3097\n",
- "\t 64687197 64687200\n",
- "{'Q.START': 3093, 'Q.END': 3094, 'T.START': 64687195, 'T.END': 64687196, 'CG': '1='}\n",
- "7046698 TO1000#1#chr03 3097 3140\n",
- "\t 64687201 64687244\n",
- "{'Q.START': 3094, 'Q.END': 3097, 'T.START': 64687197, 'T.END': 64687200, 'CG': '3='}\n",
- "7046700 TO1000#1#chr03 3140 3210\n",
- "\t 64687245 64687315\n",
- "{'Q.START': 3097, 'Q.END': 3140, 'T.START': 64687201, 'T.END': 64687244, 'CG': '43='}\n",
- "7046702 TO1000#1#chr03 3210 3211\n",
- "\t 64687316 64687317\n",
- "{'Q.START': 3140, 'Q.END': 3210, 'T.START': 64687245, 'T.END': 64687315, 'CG': '70='}\n",
- "7046703 TO1000#1#chr03 3211 3229\n",
- "\t 64687318 64687336\n",
- "{'Q.START': 3210, 'Q.END': 3211, 'T.START': 64687316, 'T.END': 64687317, 'CG': '1='}\n",
- "7046704 TO1000#1#chr03 3229 3230\n",
- "\t 64687337 64687338\n",
- "{'Q.START': 3211, 'Q.END': 3229, 'T.START': 64687318, 'T.END': 64687336, 'CG': '18='}\n",
- "7046706 TO1000#1#chr03 3230 3276\n",
- "\t 64687339 64687385\n",
- "{'Q.START': 3229, 'Q.END': 3230, 'T.START': 64687337, 'T.END': 64687338, 'CG': '1='}\n",
- "7046707 TO1000#1#chr03 3276 3277\n",
- "\t 64687386 64687387\n",
- "{'Q.START': 3230, 'Q.END': 3276, 'T.START': 64687339, 'T.END': 64687385, 'CG': '46='}\n",
- "7046709 TO1000#1#chr03 3277 3315\n",
- "\t 64687388 64687426\n",
- "{'Q.START': 3276, 'Q.END': 3277, 'T.START': 64687386, 'T.END': 64687387, 'CG': '1='}\n",
- "7046710 TO1000#1#chr03 3315 3316\n",
- "\t 64687427 64687428\n",
- "{'Q.START': 3277, 'Q.END': 3315, 'T.START': 64687388, 'T.END': 64687426, 'CG': '38='}\n",
- "7046712 TO1000#1#chr03 3316 3322\n",
- "\t 64687429 64687435\n",
- "{'Q.START': 3315, 'Q.END': 3316, 'T.START': 64687427, 'T.END': 64687428, 'CG': '1='}\n",
- "7046713 TO1000#1#chr03 3322 3323\n",
- "\t 64687436 64687437\n",
- "{'Q.START': 3316, 'Q.END': 3322, 'T.START': 64687429, 'T.END': 64687435, 'CG': '6='}\n",
- "7046715 TO1000#1#chr03 3323 3348\n",
- "\t 64687438 64687463\n",
- "{'Q.START': 3322, 'Q.END': 3323, 'T.START': 64687436, 'T.END': 64687437, 'CG': '1='}\n",
- "7046718 TO1000#1#chr03 3352 3353\n",
- "\t 64687464 64687465\n",
- "{'Q.START': 3323, 'Q.END': 3348, 'T.START': 64687438, 'T.END': 64687463, 'CG': '25='}\n",
- "7046717 TO1000#1#chr03 3351 3352\n",
- "\t 64687466 64687467\n",
- "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 64687464, 'T.END': 64687465, 'CG': '1='}\n",
- "7046718 TO1000#1#chr03 3352 3353\n",
- "\t 64687464 64687465\n",
- "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 64687466, 'T.END': 64687467, 'CG': '1='}\n",
- "7046717 TO1000#1#chr03 3351 3352\n",
- "\t 64687466 64687467\n",
- "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 64687464, 'T.END': 64687465, 'CG': '1='}\n",
- "7046718 TO1000#1#chr03 3352 3353\n",
- "\t 64687464 64687465\n",
- "{'Q.START': 3351, 'Q.END': 3352, 'T.START': 64687466, 'T.END': 64687467, 'CG': '1='}\n",
- "7046720 TO1000#1#chr03 3353 3354\n",
- "\t 64687468 64687469\n",
- "{'Q.START': 3352, 'Q.END': 3353, 'T.START': 64687464, 'T.END': 64687465, 'CG': '1='}\n",
- "7046722 TO1000#1#chr03 3354 3356\n",
- "\t 64687470 64687472\n",
- "{'Q.START': 3353, 'Q.END': 3354, 'T.START': 64687468, 'T.END': 64687469, 'CG': '1='}\n",
- "7046724 TO1000#1#chr03 3356 3357\n",
- "\t 64687473 64687474\n",
- "{'Q.START': 3354, 'Q.END': 3356, 'T.START': 64687470, 'T.END': 64687472, 'CG': '2='}\n",
- "7046725 TO1000#1#chr03 3357 3489\n",
- "\t 64687475 64687607\n",
- "{'Q.START': 3356, 'Q.END': 3357, 'T.START': 64687473, 'T.END': 64687474, 'CG': '1='}\n",
- "7046727 TO1000#1#chr03 3489 3490\n",
- "\t 64687608 64687609\n",
- "{'Q.START': 3357, 'Q.END': 3489, 'T.START': 64687475, 'T.END': 64687607, 'CG': '132='}\n",
- "7046728 TO1000#1#chr03 3490 3642\n",
- "\t 64687610 64687762\n",
- "{'Q.START': 3489, 'Q.END': 3490, 'T.START': 64687608, 'T.END': 64687609, 'CG': '1='}\n",
- "7046729 TO1000#1#chr03 3642 3644\n",
- "\t 64687763 64687765\n",
- "{'Q.START': 3490, 'Q.END': 3642, 'T.START': 64687610, 'T.END': 64687762, 'CG': '152='}\n",
- "7046730 TO1000#1#chr03 3644 3685\n",
- "\t 64687766 64687807\n",
- "{'Q.START': 3642, 'Q.END': 3644, 'T.START': 64687763, 'T.END': 64687765, 'CG': '2='}\n",
- "7046731 TO1000#1#chr03 3685 3687\n",
- "\t 64687808 64687810\n",
- "{'Q.START': 3644, 'Q.END': 3685, 'T.START': 64687766, 'T.END': 64687807, 'CG': '41='}\n",
- "7046733 TO1000#1#chr03 3687 3693\n",
- "\t 64687811 64687817\n",
- "{'Q.START': 3685, 'Q.END': 3687, 'T.START': 64687808, 'T.END': 64687810, 'CG': '2='}\n",
- "7046735 TO1000#1#chr03 3693 3694\n",
- "\t 64687818 64687819\n",
- "{'Q.START': 3687, 'Q.END': 3693, 'T.START': 64687811, 'T.END': 64687817, 'CG': '6='}\n",
- "7046736 TO1000#1#chr03 3694 3708\n",
- "\t 64687820 64687834\n",
- "{'Q.START': 3693, 'Q.END': 3694, 'T.START': 64687818, 'T.END': 64687819, 'CG': '1='}\n",
- "7046738 TO1000#1#chr03 3720 3721\n",
- "\t 64687835 64687836\n",
- "{'Q.START': 3694, 'Q.END': 3708, 'T.START': 64687820, 'T.END': 64687834, 'CG': '14='}\n",
- "7046739 TO1000#1#chr03 3721 3722\n",
- "\t 64687837 64687838\n",
- "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 64687835, 'T.END': 64687836, 'CG': '1='}\n",
- "7046740 TO1000#1#chr03 3716 3720\n",
- "\t 64687839 64687843\n",
- "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 64687837, 'T.END': 64687838, 'CG': '1='}\n",
- "7046738 TO1000#1#chr03 3720 3721\n",
- "\t 64687835 64687836\n",
- "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 64687839, 'T.END': 64687843, 'CG': '4='}\n",
- "7046739 TO1000#1#chr03 3721 3722\n",
- "\t 64687837 64687838\n",
- "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 64687835, 'T.END': 64687836, 'CG': '1='}\n",
- "7046740 TO1000#1#chr03 3716 3720\n",
- "\t 64687839 64687843\n",
- "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 64687837, 'T.END': 64687838, 'CG': '1='}\n",
- "7046738 TO1000#1#chr03 3720 3721\n",
- "\t 64687835 64687836\n",
- "{'Q.START': 3716, 'Q.END': 3720, 'T.START': 64687839, 'T.END': 64687843, 'CG': '4='}\n",
- "7046739 TO1000#1#chr03 3721 3722\n",
- "\t 64687837 64687838\n",
- "{'Q.START': 3720, 'Q.END': 3721, 'T.START': 64687835, 'T.END': 64687836, 'CG': '1='}\n",
- "7046741 TO1000#1#chr03 3722 3735\n",
- "\t 64687844 64687877\n",
- "{'Q.START': 3721, 'Q.END': 3722, 'T.START': 64687837, 'T.END': 64687838, 'CG': '1='}\n",
- "ALN_2\n",
- "7594382 D134#1#chr03 0 1\n",
- "\tIn path\n",
- "\t 70220037 70220038\n",
- "skipped\n",
- "\n",
- "7594369 D134#1#chr03 32 33\n",
- "\tIn path\n",
- "\t 70219216 70219217\n",
- "{'Q.START': 0, 'Q.END': 1, 'T.START': 70220037, 'T.END': 70220038, 'CG': '1='}\n",
- "7594371 D134#1#chr03 15 16\n",
- "\tIn path\n",
- "\t 70221163 70221164\n",
- "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 15, 'Q.END': 16, 'T.START': 70221163, 'T.END': 70221164, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594356 D134#1#chr03 66 67\n",
- "\tIn path\n",
- "\t 70219570 70219571\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594375 D134#1#chr03 68 69\n",
- "\tIn path\n",
- "\t 70221598 70221599\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594626 D134#1#chr03 10 11\n",
- "\tIn path\n",
- "\t 70219214 70219215\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
- "7594011 D134#1#chr03 11 12\n",
- "\tIn path\n",
- "\t 70219995 70219996\n",
- "{'Q.START': 10, 'Q.END': 11, 'T.START': 70219214, 'T.END': 70219215, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 11, 'Q.END': 12, 'T.START': 70219995, 'T.END': 70219996, 'CG': '1='}\n",
- "7594375 D134#1#chr03 68 69\n",
- "\tIn path\n",
- "\t 70221598 70221599\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594369 D134#1#chr03 32 33\n",
- "\tIn path\n",
- "\t 70219216 70219217\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
- "7594371 D134#1#chr03 15 16\n",
- "\tIn path\n",
- "\t 70221163 70221164\n",
- "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 15, 'Q.END': 16, 'T.START': 70221163, 'T.END': 70221164, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594241 D134#1#chr03 20 21\n",
- "\tIn path\n",
- "\t 70219220 70219221\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594248 D134#1#chr03 21 22\n",
- "\tNot in path\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 20, 'Q.END': 21, 'T.START': 70219220, 'T.END': 70219221, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594315 D134#1#chr03 53 54\n",
- "\tIn path\n",
- "\t 70219857 70219858\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
- "7594330 D134#1#chr03 26 27\n",
- "\tNot in path\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594315 D134#1#chr03 53 54\n",
- "\tIn path\n",
- "\t 70219857 70219858\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594369 D134#1#chr03 32 33\n",
- "\tIn path\n",
- "\t 70219216 70219217\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 32, 'Q.END': 33, 'T.START': 70219216, 'T.END': 70219217, 'CG': '1='}\n",
- "7594026 D134#1#chr03 37 38\n",
- "\tIn path\n",
- "\t 70220249 70220250\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 37, 'Q.END': 38, 'T.START': 70220249, 'T.END': 70220250, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594026 D134#1#chr03 37 38\n",
- "\tIn path\n",
- "\t 70220249 70220250\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 37, 'Q.END': 38, 'T.START': 70220249, 'T.END': 70220250, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594315 D134#1#chr03 53 54\n",
- "\tIn path\n",
- "\t 70219857 70219858\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 53, 'Q.END': 54, 'T.START': 70219857, 'T.END': 70219858, 'CG': '1='}\n",
- "7594311 D134#1#chr03 55 56\n",
- "\tIn path\n",
- "\t 70219351 70219352\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 70219351, 'T.END': 70219352, 'CG': '1='}\n",
- "7594021 D134#1#chr03 57 58\n",
- "\tIn path\n",
- "\t 70219218 70219219\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 70219218, 'T.END': 70219219, 'CG': '1X'}\n",
- "7594286 D134#1#chr03 59 60\n",
- "\tIn path\n",
- "\t 70219349 70219350\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 70219349, 'T.END': 70219350, 'CG': '1='}\n",
- "7594356 D134#1#chr03 66 67\n",
- "\tIn path\n",
- "\t 70219570 70219571\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594375 D134#1#chr03 68 69\n",
- "\tIn path\n",
- "\t 70221598 70221599\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
- "7594356 D134#1#chr03 66 67\n",
- "\tIn path\n",
- "\t 70219570 70219571\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 66, 'Q.END': 67, 'T.START': 70219570, 'T.END': 70219571, 'CG': '1='}\n",
- "7594375 D134#1#chr03 68 69\n",
- "\tIn path\n",
- "\t 70221598 70221599\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594374 D134#1#chr03 69 70\n",
- "\tIn path\n",
- "\t 70219092 70219093\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 70221598, 'T.END': 70221599, 'CG': '1='}\n",
- "7594350 D134#1#chr03 70 71\n",
- "\tIn path\n",
- "\t 70219226 70219227\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 70219092, 'T.END': 70219093, 'CG': '1='}\n",
- "7594264 D134#1#chr03 71 72\n",
- "\tIn path\n",
- "\t 70219228 70219229\n",
- "{'Q.START': 70, 'Q.END': 71, 'T.START': 70219226, 'T.END': 70219227, 'CG': '1='}\n",
- "7594207 D134#1#chr03 72 73\n",
- "\tIn path\n",
- "\t 70219230 70219231\n",
- "{'Q.START': 71, 'Q.END': 72, 'T.START': 70219228, 'T.END': 70219229, 'CG': '1='}\n",
- "7594225 D134#1#chr03 73 74\n",
- "\tIn path\n",
- "\t 70219232 70219233\n",
- "{'Q.START': 72, 'Q.END': 73, 'T.START': 70219230, 'T.END': 70219231, 'CG': '1='}\n",
- "7594227 D134#1#chr03 74 75\n",
- "\tIn path\n",
- "\t 70220150 70220151\n",
- "{'Q.START': 73, 'Q.END': 74, 'T.START': 70219232, 'T.END': 70219233, 'CG': '1='}\n",
- "7594120 D134#1#chr03 75 76\n",
- "\tIn path\n",
- "\t 70219236 70219237\n",
- "{'Q.START': 74, 'Q.END': 75, 'T.START': 70220150, 'T.END': 70220151, 'CG': '1='}\n",
- "7594132 D134#1#chr03 76 77\n",
- "\tIn path\n",
- "\t 70219777 70219778\n",
- "{'Q.START': 75, 'Q.END': 76, 'T.START': 70219236, 'T.END': 70219237, 'CG': '1='}\n",
- "7594165 D134#1#chr03 77 78\n",
- "\tIn path\n",
- "\t 70219240 70219241\n",
- "{'Q.START': 76, 'Q.END': 77, 'T.START': 70219777, 'T.END': 70219778, 'CG': '1='}\n",
- "7594172 D134#1#chr03 78 3735\n",
- "\tNot in path\n",
- "ALN_2\n",
- "7594382 TO1000#1#chr03 0 1\n",
- "\t 61731222 61731223\n",
- "skipped\n",
- "\n",
- "7594369 TO1000#1#chr03 32 33\n",
- "\t 61731060 61731061\n",
- "{'Q.START': 0, 'Q.END': 1, 'T.START': 61731222, 'T.END': 61731223, 'CG': '1='}\n",
- "7594371 TO1000#1#chr03 15 16\n",
- "\tNot in path\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 32, 'Q.END': 33, 'T.START': 61731060, 'T.END': 61731061, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594356 TO1000#1#chr03 66 67\n",
- "\t 61731519 61731520\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 66, 'Q.END': 67, 'T.START': 61731519, 'T.END': 61731520, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594375 TO1000#1#chr03 68 69\n",
- "\t 61733612 61733613\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594626 TO1000#1#chr03 10 11\n",
- "\t 61731056 61731057\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
- "7594011 TO1000#1#chr03 11 12\n",
- "\t 61733900 61733901\n",
- "{'Q.START': 10, 'Q.END': 11, 'T.START': 61731056, 'T.END': 61731057, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 11, 'Q.END': 12, 'T.START': 61733900, 'T.END': 61733901, 'CG': '1='}\n",
- "7594375 TO1000#1#chr03 68 69\n",
- "\t 61733612 61733613\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594369 TO1000#1#chr03 32 33\n",
- "\t 61731060 61731061\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
- "7594371 TO1000#1#chr03 15 16\n",
- "\tNot in path\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 32, 'Q.END': 33, 'T.START': 61731060, 'T.END': 61731061, 'CG': '1='}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594241 TO1000#1#chr03 20 21\n",
- "\t 61731046 61731047\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594248 TO1000#1#chr03 21 22\n",
- "\t 61734261 61734262\n",
- "{'Q.START': 20, 'Q.END': 21, 'T.START': 61731046, 'T.END': 61731047, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 21, 'Q.END': 22, 'T.START': 61734261, 'T.END': 61734262, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594315 TO1000#1#chr03 53 54\n",
- "\t 61733937 61733938\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 53, 'Q.END': 54, 'T.START': 61733937, 'T.END': 61733938, 'CG': '1='}\n",
- "7594330 TO1000#1#chr03 26 27\n",
- "\t 61731768 61731769\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 26, 'Q.END': 27, 'T.START': 61731768, 'T.END': 61731769, 'CG': '1='}\n",
- "7594315 TO1000#1#chr03 53 54\n",
- "\t 61733937 61733938\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 53, 'Q.END': 54, 'T.START': 61733937, 'T.END': 61733938, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594369 TO1000#1#chr03 32 33\n",
- "\t 61731060 61731061\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 32, 'Q.END': 33, 'T.START': 61731060, 'T.END': 61731061, 'CG': '1='}\n",
- "7594026 TO1000#1#chr03 37 38\n",
- "\t 61734267 61734268\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 37, 'Q.END': 38, 'T.START': 61734267, 'T.END': 61734268, 'CG': '1='}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594026 TO1000#1#chr03 37 38\n",
- "\t 61734267 61734268\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 37, 'Q.END': 38, 'T.START': 61734267, 'T.END': 61734268, 'CG': '1='}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594315 TO1000#1#chr03 53 54\n",
- "\t 61733937 61733938\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 53, 'Q.END': 54, 'T.START': 61733937, 'T.END': 61733938, 'CG': '1='}\n",
- "7594311 TO1000#1#chr03 55 56\n",
- "\t 61731052 61731053\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 55, 'Q.END': 56, 'T.START': 61731052, 'T.END': 61731053, 'CG': '1='}\n",
- "7594021 TO1000#1#chr03 57 58\n",
- "\t 61730922 61730923\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 57, 'Q.END': 58, 'T.START': 61730922, 'T.END': 61730923, 'CG': '1X'}\n",
- "7594286 TO1000#1#chr03 59 60\n",
- "\t 61731054 61731055\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 59, 'Q.END': 60, 'T.START': 61731054, 'T.END': 61731055, 'CG': '1='}\n",
- "7594356 TO1000#1#chr03 66 67\n",
- "\t 61731519 61731520\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 66, 'Q.END': 67, 'T.START': 61731519, 'T.END': 61731520, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594375 TO1000#1#chr03 68 69\n",
- "\t 61733612 61733613\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
- "7594356 TO1000#1#chr03 66 67\n",
- "\t 61731519 61731520\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 66, 'Q.END': 67, 'T.START': 61731519, 'T.END': 61731520, 'CG': '1='}\n",
- "7594375 TO1000#1#chr03 68 69\n",
- "\t 61733612 61733613\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594374 TO1000#1#chr03 69 70\n",
- "\t 61730920 61730921\n",
- "{'Q.START': 68, 'Q.END': 69, 'T.START': 61733612, 'T.END': 61733613, 'CG': '1='}\n",
- "7594350 TO1000#1#chr03 70 71\n",
- "\t 61731066 61731067\n",
- "{'Q.START': 69, 'Q.END': 70, 'T.START': 61730920, 'T.END': 61730921, 'CG': '1='}\n",
- "7594264 TO1000#1#chr03 71 72\n",
- "\t 61731068 61731069\n",
- "{'Q.START': 70, 'Q.END': 71, 'T.START': 61731066, 'T.END': 61731067, 'CG': '1='}\n",
- "7594207 TO1000#1#chr03 72 73\n",
- "\t 61731070 61731071\n",
- "{'Q.START': 71, 'Q.END': 72, 'T.START': 61731068, 'T.END': 61731069, 'CG': '1='}\n",
- "7594225 TO1000#1#chr03 73 74\n",
- "\t 61731072 61731073\n",
- "{'Q.START': 72, 'Q.END': 73, 'T.START': 61731070, 'T.END': 61731071, 'CG': '1='}\n",
- "7594227 TO1000#1#chr03 74 75\n",
- "\tNot in path\n",
- "7594120 TO1000#1#chr03 75 76\n",
- "\t 61731076 61731077\n",
- "{'Q.START': 73, 'Q.END': 74, 'T.START': 61731072, 'T.END': 61731073, 'CG': '1='}\n",
- "7594132 TO1000#1#chr03 76 77\n",
- "\t 61733800 61733801\n",
- "{'Q.START': 75, 'Q.END': 76, 'T.START': 61731076, 'T.END': 61731077, 'CG': '1='}\n",
- "7594165 TO1000#1#chr03 77 78\n",
- "\t 61731080 61731081\n",
- "{'Q.START': 76, 'Q.END': 77, 'T.START': 61733800, 'T.END': 61733801, 'CG': '1='}\n",
- "7594172 TO1000#1#chr03 78 3735\n",
- "\tNot in path\n"
- ]
- }
- ],
- "source": [
- "ALNS = {}\n",
- "## Iterating over alignments\n",
- "for aln_name in aln_dict.keys():\n",
- " \n",
- " ## Iterating over paths of the gfa\n",
- " for path_name in paths.keys():\n",
- " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(aln_name)\n",
- " _ = [] # Temporary list holding alignment blocks\n",
- "\n",
- " ## Iterating over alignment nodes of the current alignment\n",
- " for node_id, orient in aln_dict[aln_name][\"PATH.MATCH\"]:\n",
- "\n",
- " # Getting node info\n",
- " n_info = nodes[node_id]\n",
- " q_start = n_info[aln_name][\"START\"] # Start position on the query\n",
- " q_end = n_info[aln_name][\"END\"] # End position on the query\n",
- " _CG = n_info[aln_name][\"CIGAR\"] # Cigar of the alignment on the current node\n",
- "\n",
- " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(node_id, path_name, q_start, q_end)\n",
- "\n",
- " ## Checking if path is traversing the current node\n",
- " if path_name in list(n_info.keys()):\n",
- " if path_name == \"D134#1#chr03\": print(\"\\tIn path\")\n",
- "\n",
- " ## Getting start and end position on the target given the orientation of the node in the alignment and the path\n",
- " if n_info[aln_name][\"STRAND\"] == n_info[path_name][\"STRAND\"] :\n",
- " t_start = n_info[path_name][\"START\"]+n_info[aln_name][\"S.OFF\"]\n",
- " t_end = n_info[path_name][\"END\"]+n_info[aln_name][\"E.OFF\"] \n",
- " else :\n",
- " t_end = n_info[path_name][\"START\"]+n_info[aln_name][\"S.OFF\"]\n",
- " t_start = n_info[path_name][\"END\"]+n_info[aln_name][\"E.OFF\"]\n",
- "\n",
- " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"\\t\", t_start, t_end)\n",
- "\n",
- " \"\"\"\n",
- " If the latest block t.end and q.end matches with the current node t.start and q.start, \n",
- " the node should be added to the block. Else, we terminate the block and add the node to a new block\n",
- " \"\"\"\n",
- " \n",
- " # Non empty temporary list of aln and ending of the last block is the same as the start of the new node : \n",
- " if len(_) and _[-1][\"T.END\"] == t_start and _[-1][\"Q.END\"]+1 == q_start: \n",
- " tmp_aln[\"Q.END\"] = q_end\n",
- " tmp_aln[\"T.END\"] = t_end\n",
- " tmp_aln[\"CG\"] += _CG\n",
- "# elif len(_) and _[-1][\"T.END\"] == t_start: # Following on the target not on the query (i.e. Insertion)\n",
- "# tmp_aln[\"T.END\"] = t_end\n",
- "# tmp_aln[\"CG\"] += f\"{nodes_length[node_id]}I\"\n",
- "# elif len(_) and _[-1][\"Q.END\"]+1 == q_start: # Following on the query, not on the target (i.e. Deletion)\n",
- "# tmp_aln[\"Q.END\"] = q_end\n",
- "# tmp_aln[\"CG\"] += f\"{nodes_length[node_id]}D\"\n",
- " else : # Else, completely different\n",
- " try : \n",
- " _.append(tmp_aln)\n",
- " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(tmp_aln)\n",
- " except : \n",
- " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"skipped\\n\")\n",
- " tmp_aln = {\n",
- " \"Q.START\": q_start,\n",
- " \"Q.END\": q_end,\n",
- " \"T.START\": t_start,\n",
- " \"T.END\": t_end,\n",
- " \"CG\": _CG,\n",
- " }\n",
- " \n",
- " else : \n",
- " if path_name in [\"TO1000#1#chr03\", \"D134#1#chr03\"]: print(\"\\tNot in path\")\n",
- " # Node is not in the path\n",
- "\n",
- " del tmp_aln\n",
- " \n",
- " ALNS[(path_name, aln_name)] = _"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 9,
- "id": "547f03fa-cbd5-42f9-b668-1ca4404795ba",
- "metadata": {},
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "[{'Q.START': 0, 'Q.END': 77, 'T.START': 73306158, 'T.END': 73306235, 'CG': '77='}, {'Q.START': 77, 'Q.END': 82, 'T.START': 73306238, 'T.END': 73306243, 'CG': '5='}, {'Q.START': 83, 'Q.END': 138, 'T.START': 73306246, 'T.END': 73306301, 'CG': '55='}, {'Q.START': 139, 'Q.END': 202, 'T.START': 73306302, 'T.END': 73306365, 'CG': '63='}, {'Q.START': 202, 'Q.END': 203, 'T.START': 73306366, 'T.END': 73306367, 'CG': '1='}, {'Q.START': 203, 'Q.END': 379, 'T.START': 73306368, 'T.END': 73306544, 'CG': '176='}, {'Q.START': 379, 'Q.END': 380, 'T.START': 73306545, 'T.END': 73306546, 'CG': '1='}, {'Q.START': 380, 'Q.END': 429, 'T.START': 73306547, 'T.END': 73306596, 'CG': '49='}, {'Q.START': 429, 'Q.END': 430, 'T.START': 73306597, 'T.END': 73306598, 'CG': '1='}, {'Q.START': 430, 'Q.END': 457, 'T.START': 73306599, 'T.END': 73306626, 'CG': '27='}, {'Q.START': 457, 'Q.END': 492, 'T.START': 73306641, 'T.END': 73306676, 'CG': '35='}, {'Q.START': 508, 'Q.END': 564, 'T.START': 73306694, 'T.END': 73306750, 'CG': '56='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 568, 'Q.END': 569, 'T.START': 73306753, 'T.END': 73306754, 'CG': '1='}, {'Q.START': 569, 'Q.END': 824, 'T.START': 73306755, 'T.END': 73307010, 'CG': '255='}, {'Q.START': 826, 'Q.END': 858, 'T.START': 73307011, 'T.END': 73307043, 'CG': '32='}, {'Q.START': 858, 'Q.END': 859, 'T.START': 73307044, 'T.END': 73307045, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 868, 'Q.END': 869, 'T.START': 73307046, 'T.END': 73307047, 'CG': '1='}, {'Q.START': 869, 'Q.END': 913, 'T.START': 73307048, 'T.END': 73307092, 'CG': '44='}, {'Q.START': 913, 'Q.END': 919, 'T.START': 73307093, 'T.END': 73307099, 'CG': '6='}, {'Q.START': 919, 'Q.END': 978, 'T.START': 73307100, 'T.END': 73307159, 'CG': '59='}, {'Q.START': 978, 'Q.END': 979, 'T.START': 73307160, 'T.END': 73307161, 'CG': '1='}, {'Q.START': 979, 'Q.END': 1038, 'T.START': 73307162, 'T.END': 73307221, 'CG': '59='}, {'Q.START': 1038, 'Q.END': 1045, 'T.START': 73307224, 'T.END': 73307231, 'CG': '7='}, {'Q.START': 1045, 'Q.END': 1046, 'T.START': 73307232, 'T.END': 73307233, 'CG': '1='}, {'Q.START': 1046, 'Q.END': 1080, 'T.START': 73307234, 'T.END': 73307268, 'CG': '34='}, {'Q.START': 1080, 'Q.END': 1081, 'T.START': 73307269, 'T.END': 73307270, 'CG': '1='}, {'Q.START': 1081, 'Q.END': 1107, 'T.START': 73307271, 'T.END': 73307297, 'CG': '26='}, {'Q.START': 1108, 'Q.END': 1183, 'T.START': 73307300, 'T.END': 73307375, 'CG': '75='}, {'Q.START': 1183, 'Q.END': 1186, 'T.START': 73307376, 'T.END': 73307379, 'CG': '3='}, {'Q.START': 1224, 'Q.END': 1257, 'T.START': 73307419, 'T.END': 73307452, 'CG': '33='}, {'Q.START': 1289, 'Q.END': 1311, 'T.START': 73307475, 'T.END': 73307497, 'CG': '22='}, {'Q.START': 1359, 'Q.END': 1382, 'T.START': 73307546, 'T.END': 73307569, 'CG': '23='}, {'Q.START': 1434, 'Q.END': 1451, 'T.START': 73307643, 'T.END': 73307660, 'CG': '17='}, {'Q.START': 1451, 'Q.END': 1531, 'T.START': 73307661, 'T.END': 73307741, 'CG': '80='}, {'Q.START': 1532, 'Q.END': 1543, 'T.START': 73307744, 'T.END': 73307755, 'CG': '11='}, {'Q.START': 1544, 'Q.END': 1572, 'T.START': 73307758, 'T.END': 73307786, 'CG': '28='}, {'Q.START': 1572, 'Q.END': 1573, 'T.START': 73307787, 'T.END': 73307788, 'CG': '1='}, {'Q.START': 1573, 'Q.END': 1587, 'T.START': 73307789, 'T.END': 73307803, 'CG': '14='}, {'Q.START': 1588, 'Q.END': 1616, 'T.START': 73307806, 'T.END': 73307834, 'CG': '28='}, {'Q.START': 1616, 'Q.END': 1617, 'T.START': 73307835, 'T.END': 73307836, 'CG': '1='}, {'Q.START': 1617, 'Q.END': 1646, 'T.START': 73307837, 'T.END': 73307866, 'CG': '29='}, {'Q.START': 1646, 'Q.END': 1661, 'T.START': 73307867, 'T.END': 73307882, 'CG': '15='}, {'Q.START': 1661, 'Q.END': 1673, 'T.START': 73307883, 'T.END': 73307895, 'CG': '12='}, {'Q.START': 1673, 'Q.END': 1674, 'T.START': 73307896, 'T.END': 73307897, 'CG': '1='}, {'Q.START': 1674, 'Q.END': 1726, 'T.START': 73307898, 'T.END': 73307950, 'CG': '52='}, {'Q.START': 1727, 'Q.END': 1762, 'T.START': 73307953, 'T.END': 73307988, 'CG': '35='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1765, 'Q.END': 1766, 'T.START': 73307993, 'T.END': 73307994, 'CG': '1='}, {'Q.START': 1766, 'Q.END': 1767, 'T.START': 73307991, 'T.END': 73307992, 'CG': '1='}, {'Q.START': 1767, 'Q.END': 1824, 'T.START': 73307995, 'T.END': 73308052, 'CG': '57='}, {'Q.START': 1824, 'Q.END': 1825, 'T.START': 73308053, 'T.END': 73308054, 'CG': '1='}, {'Q.START': 1825, 'Q.END': 1975, 'T.START': 73308055, 'T.END': 73308205, 'CG': '150='}, {'Q.START': 1976, 'Q.END': 2015, 'T.START': 73308208, 'T.END': 73308247, 'CG': '39='}, {'Q.START': 2016, 'Q.END': 2047, 'T.START': 73308250, 'T.END': 73308281, 'CG': '31='}, {'Q.START': 2047, 'Q.END': 2055, 'T.START': 73308286, 'T.END': 73308294, 'CG': '8='}, {'Q.START': 2056, 'Q.END': 2120, 'T.START': 73308297, 'T.END': 73308361, 'CG': '64='}, {'Q.START': 2120, 'Q.END': 2121, 'T.START': 73308362, 'T.END': 73308363, 'CG': '1='}, {'Q.START': 2121, 'Q.END': 2157, 'T.START': 73308364, 'T.END': 73308400, 'CG': '36='}, {'Q.START': 2158, 'Q.END': 2170, 'T.START': 73308403, 'T.END': 73308415, 'CG': '12='}, {'Q.START': 2170, 'Q.END': 2171, 'T.START': 73308416, 'T.END': 73308417, 'CG': '1='}, {'Q.START': 2171, 'Q.END': 2205, 'T.START': 73308418, 'T.END': 73308452, 'CG': '34='}, {'Q.START': 2206, 'Q.END': 2344, 'T.START': 73308455, 'T.END': 73308593, 'CG': '138='}, {'Q.START': 2345, 'Q.END': 2364, 'T.START': 73308596, 'T.END': 73308615, 'CG': '19='}, {'Q.START': 2364, 'Q.END': 2383, 'T.START': 73308616, 'T.END': 73308635, 'CG': '19='}, {'Q.START': 2383, 'Q.END': 2408, 'T.START': 73308636, 'T.END': 73308661, 'CG': '25='}, {'Q.START': 2408, 'Q.END': 2409, 'T.START': 73308662, 'T.END': 73308663, 'CG': '1='}, {'Q.START': 2409, 'Q.END': 2441, 'T.START': 73308664, 'T.END': 73308696, 'CG': '32='}, {'Q.START': 2441, 'Q.END': 2442, 'T.START': 73308697, 'T.END': 73308698, 'CG': '1='}, {'Q.START': 2442, 'Q.END': 2580, 'T.START': 73308699, 'T.END': 73308837, 'CG': '138='}, {'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}, {'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}, {'Q.START': 2582, 'Q.END': 2583, 'T.START': 73308838, 'T.END': 73308839, 'CG': '1='}, {'Q.START': 2583, 'Q.END': 2584, 'T.START': 73308840, 'T.END': 73308841, 'CG': '1='}, {'Q.START': 2584, 'Q.END': 2764, 'T.START': 73308842, 'T.END': 73309022, 'CG': '180='}, {'Q.START': 2765, 'Q.END': 2797, 'T.START': 73309025, 'T.END': 73309057, 'CG': '32='}, {'Q.START': 2798, 'Q.END': 2878, 'T.START': 73309060, 'T.END': 73309140, 'CG': '80='}, {'Q.START': 2878, 'Q.END': 2879, 'T.START': 73309141, 'T.END': 73309142, 'CG': '1='}, {'Q.START': 2879, 'Q.END': 2951, 'T.START': 73309143, 'T.END': 73309215, 'CG': '72='}, {'Q.START': 2951, 'Q.END': 2952, 'T.START': 73309216, 'T.END': 73309217, 'CG': '1='}, {'Q.START': 2952, 'Q.END': 3002, 'T.START': 73309218, 'T.END': 73309268, 'CG': '50='}, {'Q.START': 3002, 'Q.END': 3077, 'T.START': 73309271, 'T.END': 73309346, 'CG': '75='}, {'Q.START': 3077, 'Q.END': 3078, 'T.START': 73309347, 'T.END': 73309348, 'CG': '1='}, {'Q.START': 3078, 'Q.END': 3093, 'T.START': 73309349, 'T.END': 73309364, 'CG': '15='}, {'Q.START': 3094, 'Q.END': 3097, 'T.START': 73309367, 'T.END': 73309370, 'CG': '3='}, {'Q.START': 3097, 'Q.END': 3140, 'T.START': 73309371, 'T.END': 73309414, 'CG': '43='}, {'Q.START': 3140, 'Q.END': 3210, 'T.START': 73309415, 'T.END': 73309485, 'CG': '70='}, {'Q.START': 3210, 'Q.END': 3211, 'T.START': 73309486, 'T.END': 73309487, 'CG': '1='}, {'Q.START': 3211, 'Q.END': 3229, 'T.START': 73309488, 'T.END': 73309506, 'CG': '18='}, {'Q.START': 3229, 'Q.END': 3230, 'T.START': 73309507, 'T.END': 73309508, 'CG': '1='}, {'Q.START': 3230, 'Q.END': 3276, 'T.START': 73309509, 'T.END': 73309555, 'CG': '46='}, {'Q.START': 3277, 'Q.END': 3315, 'T.START': 73309558, 'T.END': 73309596, 'CG': '38='}, {'Q.START': 3316, 'Q.END': 3322, 'T.START': 73309599, 'T.END': 73309605, 'CG': '6='}, {'Q.START': 3323, 'Q.END': 3348, 'T.START': 73309608, 'T.END': 73309633, 'CG': '25='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3351, 'Q.END': 3352, 'T.START': 73309636, 'T.END': 73309637, 'CG': '1='}, {'Q.START': 3352, 'Q.END': 3353, 'T.START': 73309634, 'T.END': 73309635, 'CG': '1='}, {'Q.START': 3353, 'Q.END': 3354, 'T.START': 73309638, 'T.END': 73309639, 'CG': '1='}, {'Q.START': 3354, 'Q.END': 3356, 'T.START': 73309640, 'T.END': 73309642, 'CG': '2='}, {'Q.START': 3357, 'Q.END': 3489, 'T.START': 73309645, 'T.END': 73309777, 'CG': '132='}, {'Q.START': 3490, 'Q.END': 3642, 'T.START': 73309780, 'T.END': 73309932, 'CG': '152='}, {'Q.START': 3644, 'Q.END': 3685, 'T.START': 73309933, 'T.END': 73309974, 'CG': '41='}, {'Q.START': 3687, 'Q.END': 3693, 'T.START': 73309977, 'T.END': 73309983, 'CG': '6='}, {'Q.START': 3694, 'Q.END': 3708, 'T.START': 73309986, 'T.END': 73310000, 'CG': '14='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}, {'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}, {'Q.START': 3716, 'Q.END': 3720, 'T.START': 73310005, 'T.END': 73310009, 'CG': '4='}, {'Q.START': 3720, 'Q.END': 3721, 'T.START': 73310010, 'T.END': 73310011, 'CG': '1='}, {'Q.START': 3721, 'Q.END': 3722, 'T.START': 73310003, 'T.END': 73310004, 'CG': '1='}]\n"
- ]
- }
- ],
- "source": [
- "print(ALNS[(\"D134#1#chr03\", \"ALN_1\")])"
- ]
- }
- ],
- "metadata": {
- "kernelspec": {
- "display_name": "Python 3 (ipykernel)",
- "language": "python",
- "name": "python3"
- },
- "language_info": {
- "codemirror_mode": {
- "name": "ipython",
- "version": 3
- },
- "file_extension": ".py",
- "mimetype": "text/x-python",
- "name": "python",
- "nbconvert_exporter": "python",
- "pygments_lexer": "ipython3",
- "version": "3.10.14"
- }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}
diff --git a/gaf2aln.py b/gaf2aln.py
deleted file mode 100644
index 25fdebf..0000000
--- a/gaf2aln.py
+++ /dev/null
@@ -1,441 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-"""
-gaf2aln
-Convert gaf alignement to sam or paf
-
-@author: alexis.mergez@inrae.fr
-@version: 0.1
-"""
-
-import numpy as np
-import pandas as pd
-import argparse
-import concurrent.futures
-import os
-import re
-
-version = "0.1"
-
-## Argument parser
-arg_parser = argparse.ArgumentParser(description='GFAvc: GFA version converter')
-arg_parser.add_argument(
- "--gfa",
- "-g",
- dest = "gfa",
- required = True,
- help = "Graph (.gfa v1)"
- )
-arg_parser.add_argument(
- "--gaf",
- "-a",
- dest = "gaf",
- required = True,
- help = "Alignement file (.gaf)"
- )
-arg_parser.add_argument(
- "--format",
- "-f",
- dest = "format",
- default = "P",
- help = "Output file format. (S: sam, P: paf (default))"
- )
-arg_parser.add_argument(
- "--threads",
- "-t",
- dest = "threads",
- required = False,
- default = 1,
- type = int,
- help = "Number of threads"
- )
-arg_parser.add_argument(
- '--version',
- '-v',
- action="store_true",
- dest = "version",
- help = "Show version"
-)
-args = arg_parser.parse_args()
-
-# Printing version
-if args.version:
- print(version)
- os._exit(0)
-
-# Toolbox
-def walk2path(walk):
- """
- Takes a walk in a single string and returns a list of nodes id with signs (gfa v1 like)
- """
- _ = re.findall(r'>\w+|<\w+', walk)
- # Converting ['>..', '>..', '<..', '>..'] to '..+,..+,..-,..+'
- return [f'{elem[1:]}{(elem[0] == ">")*"+"+(elem[0] == "<")*"-"}' for elem in _]
-
-def cigar2basealn(cigar):
- """
- Takes a CIGAR string and convert it into a list of base level alignment.
- For example : "345=" -> ["=", "=", ..., "="] of length 345.
- """
- _ = re.findall(r'\d+\D', cigar)
- final_cigar = []
- for match in _:
- final_cigar += [match[-1]]*int(match[:-1])
-
- return final_cigar
-
-def basealn2cigar(base_aln_list):
-
- last_elem = base_aln_list[0]
- CIGAR = [[1, last_elem]]
- for elem in base_aln_list[1:]:
- if elem == last_elem:
- CIGAR[-1][0] += 1
-
- else :
- CIGAR[-1][0] = str(CIGAR[-1][0])
- CIGAR.append([1, elem])
- last_elem = elem
- CIGAR[-1][0] = str(CIGAR[-1][0])
- return "".join(["".join(block) for block in CIGAR if block[1] != ""])
-
-# Parsing the .gaf file
-print(f"[gaf2aln::GAF Parser] Reading {args.gaf} ...")
-with open(args.gaf, 'r') as file:
- gaf_lines = file.readlines()
-
-gaf_col = [
- "QRY.NAME", "QRY.LEN", "QRY.START", "QRY.END", "STRAND",
- "PATH.MATCH", "PATH.LEN", "ALN.START", "ALN.END",
- "RES.MATCH", "ALN.BLOCK.LEN", "MAPPING.QUAL"
- ]
-
-# Creating dictionnary to store alignments
-print(f"[gaf2aln::GAF Parser] Extracting alignments ...")
-aln_dict = {}
-for line in range(len(gaf_lines)):
- ## Splitting the line by tabulation
- line_content = gaf_lines[line][:-1].split('\t')
-
- ## Adding alignement info to dictionnary
- aln_dict[f"ALN_{line+1}"] = {
- gaf_col[i]: line_content[i] for i in range(len(gaf_col))
- }
-
- ## Splitting "PATH.MATCH" into a list
- aln_dict[f"ALN_{line+1}"]["PATH.MATCH"] = [
- (str(node_id[:-1]), node_id[-1])
- for node_id in walk2path(aln_dict[f"ALN_{line+1}"]["PATH.MATCH"])
- ]
-
- ## Adding CIGAR
- aln_dict[f"ALN_{line+1}"]["RAW.CIGAR"] = line_content[-1]
-
- ## Adding tags
- aln_dict[f"ALN_{line+1}"]["TAGS"] = ",".join(line_content[13:-1])
-
-# Getting nodes of interest ids
-aln_nodes = np.unique([
- str(node_id)
- for aln in aln_dict.keys()
- for node_id, orient in aln_dict[aln]["PATH.MATCH"]
-]).tolist()
-
-del gaf_lines, gaf_col
-
-# Parsing the .gfa
-print(f"[gaf2aln::GFA Parser] Reading {args.gfa} ...")
-with open(args.gfa, 'r') as file:
- gfa_lines = file.readlines()
-
-# Nodes length dictionnary structured as follow :
-# {<NODE.ID>: <NODE.LENGTH>}
-nodes_length = {}
-# Nodes dictionnary structured as follow :
-# { <ALN.NODE.ID> : {
-# <PATH.NAME>: {"START": start, "END": end, "STRAND": strand),
-# <ALN.NAME>: {"START": start, "END": end, "S.OFF": start.offset, "E.OFF": end.offset, "STRAND": strand, "CIGAR": CIGAR}
-# }
-# }
-nodes = {node_id: {} for node_id in aln_nodes}
-# Paths dictionnary structured as follow :
-# {<PATH.NAME>: {NODES: {<NODE.ID>: <NODE.ORIENT>}, CIGAR: <CIGAR in comma separated list>}
-paths = {}
-# Links dictionnary structured as follow :
-# {<FROM.NODE.ID>: {<TO.NODE.ID>: {FROM.ORIENT: <FROM.ORIENT>, TO.ORIENT: <TO.ORIENT>}}}
-links = {}
-
-# Parsing the gfa
-print(f"[gaf2aln::GFA Parser] Extracting nodes, paths and links ...")
-
-def GFA_parser(gfa_lines, nodes = nodes):
- _links, _nodes, _nodes_length, paths = {}, {}, {}, {}
- for line in gfa_lines:
- line_content = line[:-1].split("\t")
- line_id = line_content[0]
-
- # Segment line
- if line_id == "S" :
-
- _nodes_length[str(line_content[1])] = len(line_content[2])
-
- # Link line
- elif line_id == "L":
- try :
- _links[str(line_content[1])][str(line_content[3])] = {
- "FROM": str(line_content[2]),
- "TO": str(line_content[4])
- }
-
- except :
- _links[str(line_content[1])] = {
- str(line_content[3]) : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
- }
-
- # Path line
- elif line_id == "P":
- _paths[str(line_content[1])] = {
- "NODES": {
- str(node_id[:-1]): str(node_id[-1])
- for node_id in line_content[2].split(',')
- },
- "CIGAR": line_content[3]
- }
-
- return [_links, _nodes, _nodes_length, _paths]
-
-# splits = np.quantile(range(len(gfa_lines)+1), q= np.array(args.threads+1)/args.threads, method='higher').tolist()
-# res = []
-# for i in range(1, len(splits)):
-# res.append(executor.submit(GFA_parser, gfa_lines[splits[i-1]:splits[i]]))
-
-# for out in res:
-# results = out.result()
-
-# for link_id, link_info in results[0].items():
-# links[]
-
-
-for line in gfa_lines:
- line_content = line[:-1].split("\t")
- line_id = line_content[0]
-
- # Segment line
- if line_id == "S" :
-
- nodes_length[str(line_content[1])] = len(line_content[2])
-
- # Link line
- elif line_id == "L":
- try :
- links[str(line_content[1])][str(line_content[3])] = {
- "FROM": str(line_content[2]),
- "TO": str(line_content[4])
- }
-
- except :
- links[str(line_content[1])] = {
- str(line_content[3]) : {"FROM.ORIENT": str(line_content[2]), "TO.ORIENT": str(line_content[4])}
- }
-
- # Path line
- elif line_id == "P":
- paths[str(line_content[1])] = {
- "NODES": {
- str(node_id[:-1]): str(node_id[-1])
- for node_id in line_content[2].split(',')
- },
- "CIGAR": line_content[3]
- }
-
-del gfa_lines
-
-print(f"[gaf2aln::Graph position processing] Computing nodes positions in each paths...")
-def get_node_pos(path_name, nodes = nodes, paths = paths, nodes_length = nodes_length):
- print(f"[gaf2aln::Graph position processing] Running on {path_name} ...")
- cur_pos = 0
-
- out = {}
- # Iterating over nodes in the path
- for path_node in paths[path_name]["NODES"].keys():
- # Instead of checking if the node is one interesting node, we try to add to the nodes dict
- if path_node in aln_nodes :
- out[path_node] = {
- "START": cur_pos, # Start position of the node start in the currrent path
- "END": cur_pos+nodes_length[path_node], # End position of the node end in the current path
- "STRAND": paths[path_name]["NODES"][path_node] # Orientation of the node in the current path
- }
-
- cur_pos += nodes_length[path_node]+1
- else :
- cur_pos += nodes_length[path_node]+1
-
- return out
-
-res = {}
-executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
-# Adding nodes positions relative to path
-for path_name in paths.keys():
- res[path_name] = executor.submit(get_node_pos, path_name)
-
-executor.shutdown(wait=True)
-
-for path_name, out in res.items():
- results = out.result()
- for path_node, node_pos in results.items():
- nodes[path_node][path_name] = node_pos
-
-del res
-
-print(f"[gaf2aln::Alignment position processing] Computing nodes positions in each alignement...")
-# Adding nodes positions relative to path
-
-def get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length):
- # Initializing current position in query
- cur_pos = 0
-
- # Getting start and end node ids
- start_end_id = (aln_dict[aln_name]["PATH.MATCH"][0][0], aln_dict[aln_name]["PATH.MATCH"][-1][0])
-
- # Creating result dictionnary
- res = {}
-
- ## Iterating over node_ids from the given alignment
- for node_id, orient in aln_dict[aln_name]["PATH.MATCH"]:
- # Adding entry for current node
- res[node_id] = {aln_name: {}}
-
- # First node
- if node_id == start_end_id[0]:
- start_pos = 0
- s_off = int(aln_dict[aln_name]["ALN.START"])
- end_pos = nodes_length[node_id]-s_off
- e_off = 0
- # End node
- elif node_id == start_end_id[1]:
- start_pos = cur_pos
- s_off = 0
- end_pos = int(aln_dict[aln_name]["QRY.END"])
- e_off = nodes_length[node_id]-(end_pos-cur_pos)
- # Node in between
- else :
- start_pos = cur_pos
- s_off, e_off = 0, 0
- end_pos = cur_pos+nodes_length[node_id]
-
- res[node_id] = {
- "START": start_pos, # Start position on the query
- "END": end_pos, # End position on the query
- "S.OFF": s_off, # Offset between the start of the alignment and the node's start
- "E.OFF": e_off, # Offset between the end of the alignment and the node's end
- "STRAND": orient # Orientation of the node in the alignment
- }
-
- cur_pos = end_pos
- print(start_pos, end_pos, s_off, e_off, orient, nodes_length[node_id], cur_pos)
-
- return res
-
-# Storing alignement
-res = {}
-executor = concurrent.futures.ThreadPoolExecutor(max_workers=args.threads)
-for aln_name in aln_dict.keys():
- print(f"[gaf2aln::Alignment position processing] Running on {aln_name} ...")
-
- res[aln_name] = executor.submit(get_aln_node_info, aln_name)
- #res[aln_name] = get_aln_node_info(aln_name, aln_dict = aln_dict, nodes_length = nodes_length)
-
-executor.shutdown(wait=True)
-
-for aln_name, node_info in res.items():
- results = node_info.result()
- for node_id, info in results.items():
- nodes[node_id][aln_name] = info
-
-del res
-
-# Calculating CIGAR for each nodes in each aln
-print(f"[gaf2aln::CIGAR processing] Computing nodes cigar from alignement ...")
-# Iterating over alignments
-for aln in aln_dict.keys():
-
- print(f"[gaf2aln::CIGAR processing] Running on {aln} ...")
- # Getting the list of base level alignement (["=", "X", ...] from "1=1X...")
- raw_cigar = cigar2basealn(aln_dict[aln]["RAW.CIGAR"])
- CIGAR={}
-
- for node_id, orient in aln_dict[aln]["PATH.MATCH"]:
-
- _cigar = basealn2cigar(raw_cigar[
- nodes[node_id][aln]["START"]:nodes[node_id][aln]["END"]
- ])
- nodes[node_id][aln]["CIGAR"] = _cigar
- #print(_cigar, nodes[node_id][aln]["START"], nodes[node_id][aln]["END"])
-
-#print(nodes)
-
-# Lifting graph alignements to haplotype alignements
-
-ALNS = {}
-for aln_name in aln_dict.keys():
-
- for path_name in paths.keys():
-
- _ = []
- for node_id, orient in aln_dict[aln_name]["PATH.MATCH"]:
-
- n_info = nodes[node_id]
- q_start = n_info[aln_name]["START"]
- q_end = n_info[aln_name]["END"]
- _CG = n_info[aln_name]["CIGAR"]
-
- print(node_id, path_name, q_start, q_end)
- if path_name in list(n_info.keys()):
- print("\tIn path")
-
- if n_info[aln_name]["STRAND"] == n_info[path_name]["STRAND"] :
- t_start = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
- t_end = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
- else :
- t_end = n_info[path_name]["START"]+n_info[aln_name]["S.OFF"]
- t_start = n_info[path_name]["END"]+n_info[aln_name]["E.OFF"]
-
- print("\t", t_start, t_end)
-
- # Non empty temporary list of aln and ending of the last block is the same as the start of the new node :
- if len(_) and _[-1]["T.END"] == t_start and _[-1]["Q.END"] == q_start:
- tmp_aln["Q.END"] = q_end
- tmp_aln["T.END"] = t_end
- tmp_aln["CG"] += _CG
- elif len(_) and _[-1]["T.END"] == t_start: # Following on the target not on the query (i.e. Insertion)
- tmp_aln["T.END"] = t_end
- tmp_aln["CG"] += f"{nodes_length[node_id]}I"
- elif len(_) and _[-1]["Q.END"] == q_start: # Following on the query, not on the target (i.e. Deletion)
- tmp_aln["Q.END"] = q_end
- tmp_aln["CG"] += f"{nodes_length[node_id]}D"
- else : # Else, completely different
- try :
- _.append(tmp_aln)
- except : pass
- tmp_aln = {
- "Q.START": q_start,
- "Q.END": q_end,
- "T.START": t_start,
- "T.END": t_end,
- "CG": _CG,
- }
-
- else :
- print("\tNot in path")
- # Node is not in the path
-
-
- ALNS[(path_name, aln_name)] = _
-
-## Debug
-for elem in ALNS.keys():
- print(elem)
-
-for key, elem in ALNS.items():
- print(key)
- print(elem)
\ No newline at end of file
--
GitLab