annotate mut2read.py @ 62:66c1245436b9 draft

planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
author mheinzl
date Wed, 17 Mar 2021 13:24:40 +0000
parents d21960b45a6b
children 6ccff403db8a
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
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
23 import sys
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
24
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
25 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
26 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
27 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
28
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 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
31 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
32 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
33 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
34 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
35 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
36 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
37 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
38 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
39 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
40 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
41 help='Output JSON file to store collected data.')
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
42 return parser
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
43
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
44
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
45 def mut2read(argv):
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
46 parser = make_argparser()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
47 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
48
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
49 file1 = args.mutFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
50 file2 = args.bamFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
51 file3 = args.familiesFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
52 outfile = args.outputFastq
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
53 json_file = args.outputJson
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
54
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
55 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
56 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
57
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
58 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
59 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
60
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
61 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
62 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
63
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
64 # 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
65 # 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
66 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
67
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
68 # get tags
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
69 tag_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
70 cvrg_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
71
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
72 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
73 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
74 stop_pos = variant.start
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
75 #chrom_stop_pos = str(chrom) + "#" + str(stop_pos)
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
76 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
77 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
78 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
79 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
80 alt = 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
81 print(alt)
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
82 chrom_stop_pos = str(chrom) + "#" + str(stop_pos) + "#" + ref + "#" + alt
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
83
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
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
85 dcs_len = []
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
86 if len(ref) == len(alt):
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
87 for pileupcolumn in bam.pileup(chrom, stop_pos - 1, stop_pos + 1, max_depth=100000000):
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
88 if pileupcolumn.reference_pos == stop_pos:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
89 count_alt = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
90 count_ref = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
91 count_indel = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
92 count_n = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
93 count_other = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
94 count_lowq = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
95 print("unfiltered reads=", pileupcolumn.n, "filtered reads=", len(pileupcolumn.pileups),
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
96 "difference= ", len(pileupcolumn.pileups) - pileupcolumn.n)
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
97 for pileupread in pileupcolumn.pileups:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
98 if not pileupread.is_del and not pileupread.is_refskip:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
99 # query position is None if is_del or is_refskip is set.
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
100 nuc = pileupread.alignment.query_sequence[pileupread.query_position]
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
101 dcs_len.append(len(pileupread.alignment.query_sequence))
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
102 if nuc == alt:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
103 count_alt += 1
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
104 tag = pileupread.alignment.query_name
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
105 if tag in tag_dict:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
106 tag_dict[tag][chrom_stop_pos] = alt
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
107 else:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
108 tag_dict[tag] = {}
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
109 tag_dict[tag][chrom_stop_pos] = alt
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
110 elif nuc == ref:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
111 count_ref += 1
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
112 elif nuc == "N":
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
113 count_n += 1
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
114 elif nuc == "lowQ":
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
115 count_lowq += 1
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
116 else:
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
117 count_other += 1
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
118 else:
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
119 count_indel += 1
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
120 dcs_median = np.median(np.array(dcs_len))
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
121 cvrg_dict[chrom_stop_pos] = (count_ref, count_alt, dcs_median)
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
122
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
123 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" %
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
124 (pileupcolumn.pos, count_ref + count_alt, count_ref, count_alt, count_other, count_n,
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
125 count_indel, count_lowq, dcs_median))
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
126 else:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
127 print("indels are currently not evaluated")
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
128 bam.close()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
129
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
130 with open(json_file, "w") as f:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
131 json.dump((tag_dict, cvrg_dict), f)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
132
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
133 # 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
134 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
135 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
136 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
137 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
138 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
139 tag = splits[0]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
140
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
141 if tag in tag_dict:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
142 str1 = splits[4]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
143 curr_seq = str1.replace("-", "")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
144 str2 = splits[5]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
145 curr_qual = str2.replace(" ", "")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
146
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
147 out.write("@" + splits[0] + "." + splits[1] + "." + splits[2] + "\n")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
148 out.write(curr_seq + "\n")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
149 out.write("+" + "\n")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
150 out.write(curr_qual + "\n")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
151
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
152
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
153 if __name__ == '__main__':
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
154 sys.exit(mut2read(sys.argv))
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
155