annotate mut2read.py @ 87:49b4a9ca7400 draft

planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
author mheinzl
date Sat, 22 Apr 2023 10:04:18 +0000
parents e46d5e377760
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
1 #!/usr/bin/env python
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
2
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
3 """mut2read.py
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
4
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
5 Author -- Gundula Povysil
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
6 Contact -- povysil@bioinf.jku.at
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
7
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
8 Takes a tabular file with mutations and a BAM file as input and prints
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
9 all tags of reads that carry the mutation to a user specified output file.
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
10 Creates fastq file of reads of tags with mutation.
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
11
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
12 ======= ========== ================= ================================
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
13 Version Date Author Description
43
d21960b45a6b planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 11
diff changeset
14 0.2.1 2019-10-27 Gundula Povysil -
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
15 ======= ========== ================= ================================
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
16
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
17 USAGE: python mut2read.py DCS_Mutations.tabular DCS.bam Aligned_Families.tabular Interesting_Reads.fastq tag_count_dict.json
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
18 """
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
19
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
20 import argparse
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
21 import json
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
22 import os
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
23 import re
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
24 import sys
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
25
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
26 import numpy as np
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
27 import pysam
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
28 from cyvcf2 import VCF
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
29
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
30
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
31 def make_argparser():
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
32 parser = argparse.ArgumentParser(description='Takes a vcf file with mutations and a BAM file as input and prints all tags of reads that carry the mutation to a user specified output file and creates a fastq file of reads of tags with mutation.')
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
33 parser.add_argument('--mutFile',
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
34 help='VCF file with DCS mutations.')
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
35 parser.add_argument('--bamFile',
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
36 help='BAM file with aligned DCS reads.')
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
37 parser.add_argument('--familiesFile',
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
38 help='TABULAR file with aligned families.')
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
39 parser.add_argument('--outputFastq',
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
40 help='Output FASTQ file of reads with mutations.')
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
41 parser.add_argument('--outputJson',
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
42 help='Output JSON file to store collected data.')
84
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
43 parser.add_argument('--refalttiers', action="store_true",
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
44 help='Store also information about the reference allele.')
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
45 return parser
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
46
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
47
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
48 def mut2read(argv):
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
49 parser = make_argparser()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
50 args = parser.parse_args(argv[1:])
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
51
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
52 file1 = args.mutFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
53 file2 = args.bamFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
54 file3 = args.familiesFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
55 outfile = args.outputFastq
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
56 json_file = args.outputJson
84
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
57 refalttiers = args.refalttiers
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
58
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
59 if os.path.isfile(file1) is False:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
60 sys.exit("Error: Could not find '{}'".format(file1))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
61
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
62 if os.path.isfile(file2) is False:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
63 sys.exit("Error: Could not find '{}'".format(file2))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
64
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
65 if os.path.isfile(file3) is False:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
66 sys.exit("Error: Could not find '{}'".format(file3))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
67
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
68 # read dcs bam file
43
d21960b45a6b planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 11
diff changeset
69 # pysam.index(file2)
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
70 bam = pysam.AlignmentFile(file2, "rb")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
71
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
72 # get tags
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
73 tag_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
74 cvrg_dict = {}
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
75 tag_dict_ref = {}
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
76
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
77 for variant in VCF(file1):
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
78 chrom = variant.CHROM
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
79 stop_pos = variant.start
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
80 ref = variant.REF
43
d21960b45a6b planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 11
diff changeset
81 if len(variant.ALT) == 0:
d21960b45a6b planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 11
diff changeset
82 continue
d21960b45a6b planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 11
diff changeset
83 else:
d21960b45a6b planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 11
diff changeset
84 alt = variant.ALT[0]
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
85 alt = alt.upper()
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
86 ref = ref.upper()
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
87 if "N" in alt: # skip indels with N in alt allele --> it is not an indel but just a mismatch at the position where the N is (checked this in IGV)
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
88 continue
7
ded0dc6a20d3 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 6
diff changeset
89 chrom_stop_pos = str(chrom) + "#" + str(stop_pos) + "#" + ref + "#" + alt
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
90 dcs_len = []
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
91 for pileupcolumn in bam.pileup(chrom, stop_pos - 1, stop_pos + 1, max_depth=100000000):
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
92 if pileupcolumn.reference_pos == stop_pos:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
93 count_alt = 0
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
94 count_ref = 0
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
95 count_indel = 0
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
96 count_n = 0
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
97 count_other = 0
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
98 count_lowq = 0
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
99 for pileupread in pileupcolumn.pileups:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
100 if not pileupread.is_refskip:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
101 if pileupread.is_del:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
102 p = pileupread.query_position_or_next
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
103 e = p + len(alt) - 1
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
104 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
105 p = pileupread.query_position
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
106 e = p + len(alt)
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
107 s = p
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
108 split_cigar = re.split('(\d+)', pileupread.alignment.cigarstring)
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
109 if len(ref) < len(alt):
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
110 if "I" in split_cigar:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
111 all_insertions = [inser_i for inser_i, ins in enumerate(split_cigar) if ins == "I"]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
112 for ai in all_insertions: # if multiple insertions in DCS
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
113 ins_index = [int(ci) for ci in split_cigar[:ai - 1] if ci.isdigit()]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
114 ins_count = split_cigar[ai - 1] # nr of insertions should match with alt allele
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
115 if "I" in split_cigar and sum(ins_index) == p + 1 and len(alt) - 1 == int(ins_count): # if position in read matches and length of insertion
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
116 nuc = pileupread.alignment.query_sequence[s:e]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
117 break
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
118 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
119 nuc = pileupread.alignment.query_sequence[s]
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
120 else:
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
121 nuc = pileupread.alignment.query_sequence[s]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
122 elif len(ref) > len(alt):
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
123 ref_positions = pileupread.alignment.get_reference_positions(full_length=True)[s:p + len(ref)]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
124 if "D" in split_cigar:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
125 all_deletions = [del_i for del_i, dele in enumerate(split_cigar) if dele == "D"]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
126 for di, ai in enumerate(all_deletions): # if multiple insertions in DCS
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
127 if di > 0: # more than 1 deletion, don't count previous deletion to position
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
128 all_deletions_mod = split_cigar[:ai - 1]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
129 prev_del_idx = [all_deletions_mod.index("D") - 1, all_deletions_mod.index("D")]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
130 split_cigar_no_prev = [ad for i, ad in enumerate(all_deletions_mod) if i not in prev_del_idx]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
131 del_index = [int(ci) for ci in split_cigar_no_prev[:ai - 1] if ci.isdigit()]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
132 else: # first deletion in read, sum all previous (mis)matches and insertions to position
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
133 del_index = [int(ci) for ci in split_cigar[:ai - 1] if ci.isdigit()]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
134 del_count = split_cigar[ai - 1] # nr of deletions should match with ref allele
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
135 if "D" in split_cigar and sum(del_index) == p + 1 and len(ref) - 1 == int(del_count):
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
136 nuc = pileupread.alignment.query_sequence[s:e]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
137 if nuc == "":
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
138 nuc = str(alt)
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
139 break
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
140 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
141 nuc = pileupread.alignment.query_sequence[s:s + len(ref)]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
142 elif len(ref_positions) < len(ref): # DCS has reference but the position is at the very end of the DCS and therefore not the full reference positions are there
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
143 nuc = pileupread.alignment.get_reference_sequence()[s:s + len(ref)]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
144 if nuc.upper() == ref[:len(nuc)]:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
145 nuc = str(ref)
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
146 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
147 nuc = pileupread.alignment.query_sequence[s:s + len(ref)]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
148 else: # SNV: query position is None if is_del or is_refskip is set.
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
149 nuc = pileupread.alignment.query_sequence[s]
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
150
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
151 nuc = nuc.upper()
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
152 tag = pileupread.alignment.query_name
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
153 if "_" in tag:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
154 tag = re.split('_', tag)[0]
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
155
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
156 if nuc == alt:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
157 count_alt += 1
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
158 if tag in tag_dict:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
159 tag_dict[tag][chrom_stop_pos] = alt
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
160 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
161 tag_dict[tag] = {}
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
162 tag_dict[tag][chrom_stop_pos] = alt
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
163 elif nuc == ref:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
164 count_ref += 1
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
165 if tag in tag_dict_ref:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
166 tag_dict_ref[tag][chrom_stop_pos] = ref
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
167 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
168 tag_dict_ref[tag] = {}
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
169 tag_dict_ref[tag][chrom_stop_pos] = ref
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
170 elif nuc == "N":
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
171 count_n += 1
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
172 elif nuc == "lowQ":
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
173 count_lowq += 1
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
174 else:
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
175 count_other += 1
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
176 dcs_len.append(len(pileupread.alignment.query_sequence))
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
177
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
178 dcs_median = np.median(np.array(dcs_len))
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
179 cvrg_dict[chrom_stop_pos] = (count_ref, count_alt, dcs_median)
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
180 print("coverage at pos %s = %s, ref = %s, alt = %s, other bases = %s, N = %s, indel = %s, low quality = %s, median length of DCS = %s\n" %
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
181 (pileupcolumn.pos, count_ref + count_alt, count_ref, count_alt, count_other, count_n,
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
182 count_indel, count_lowq, dcs_median))
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
183 bam.close()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
184
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
185 with open(json_file, "w") as f:
78
fdfe9a919ff7 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 75
diff changeset
186 json.dump((tag_dict, cvrg_dict, tag_dict_ref), f)
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
187
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
188 # create fastq from aligned reads
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
189 with open(outfile, 'w') as out:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
190 with open(file3, 'r') as families:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
191 for line in families:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
192 line = line.rstrip('\n')
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
193 splits = line.split('\t')
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
194 tag = splits[0]
84
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
195 if refalttiers is True:
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
196 if tag in tag_dict or tag in tag_dict_ref:
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
197 str1 = splits[4]
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
198 curr_seq = str1.replace("-", "")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
199 str2 = splits[5]
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
200 curr_qual = str2.replace(" ", "")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
201 out.write("@" + splits[0] + "." + splits[1] + "." + splits[2] + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
202 out.write(curr_seq + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
203 out.write("+" + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
204 out.write(curr_qual + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
205 else:
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
206 if tag in tag_dict:
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
207 str1 = splits[4]
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
208 curr_seq = str1.replace("-", "")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
209 str2 = splits[5]
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
210 curr_qual = str2.replace(" ", "")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
211 out.write("@" + splits[0] + "." + splits[1] + "." + splits[2] + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
212 out.write(curr_seq + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
213 out.write("+" + "\n")
e46d5e377760 planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
mheinzl
parents: 78
diff changeset
214 out.write(curr_qual + "\n")
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
215
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
216
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
217 if __name__ == '__main__':
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
218 sys.exit(mut2read(sys.argv))