Mercurial > repos > mheinzl > variant_analyzer2

annotate mut2sscs.py @ 90:24f166c1dba7 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8-dirty
author mheinzl
date Wed, 26 Apr 2023 18:55:52 +0000
parents fdfe9a919ff7
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 """mut2sscs.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 from DCS and a BAM file of SSCS as input
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
9 and extracts all tags of reads that carry the mutation.
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
10 Calculates statistics about number of ab/ba/duplex per 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
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
17 USAGE: python mut2sscs.py DCS_Mutations.tabular SSCS.bam SSCS_counts.json
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
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
21 from __future__ import division
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
22
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
23 import argparse
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
24 import json
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
25 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
26 import re
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
27 import sys
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
28
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
29 import numpy as np
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
30 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
31 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
32
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
33
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
34 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
35 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.')
0
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('--mutFile',
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
37 help='VCR 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
38 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
39 help='BAM file with aligned SSCS reads.')
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 SSCS counts.')
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 mut2sscs(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 sscs_counts_json = args.outputJson
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
52
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
53 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
54 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
55
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
56 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
57 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
58
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
59 # read SSCS 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
60 # 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
61 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
62
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
63 # get tags
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
64 mut_pos_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
65 ref_pos_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
66
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
67 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
68 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
69 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
70 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
71 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
72 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
73 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
74 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
75 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
76 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
77 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
78 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
79 chrom_stop_pos = str(chrom) + "#" + str(stop_pos) + "#" + ref + "#" + alt
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
80 for pileupcolumn in bam.pileup(chrom, stop_pos - 1, stop_pos + 1, max_depth=1000000000):
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
81 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
82 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
83 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
84 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
85 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
86 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
87 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
88 abba = tag[-2:]
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
89 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
90 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
91 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
92 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
93 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
94 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
95 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
96 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
97 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
98 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
99 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
100 if len(ref) < len(alt): # 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
101 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
102 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
103 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
104 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
105 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
106 if "I" in split_cigar and sum(ins_index) == p + 1 and len(alt) - 1 == int(ins_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
107 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
108 break
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
109 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
110 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
111 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
112 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
113 elif len(ref) > len(alt): # deletion
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 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
115 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
116 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
117 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
118 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
119 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
120 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
121 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
122 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
123 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
124 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
125 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
126 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
127 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
128 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
129 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
130 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
131 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
132 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
133 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
134 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
135 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
136 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
137 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
138 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
139 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
140 nuc = pileupread.alignment.query_sequence[s]
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
141
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
142 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
143
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 == 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
145 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
146 if chrom_stop_pos in mut_pos_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
147 if abba in mut_pos_dict[chrom_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
148 mut_pos_dict[chrom_stop_pos][abba] += 1
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
149 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
150 mut_pos_dict[chrom_stop_pos][abba] = 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
151 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
152 mut_pos_dict[chrom_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
153 mut_pos_dict[chrom_stop_pos][abba] = 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
154 if chrom_stop_pos not in ref_pos_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
155 ref_pos_dict[chrom_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
156 ref_pos_dict[chrom_stop_pos][abba] = 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
157 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
158 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
159 if chrom_stop_pos in ref_pos_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
160 if abba in ref_pos_dict[chrom_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
161 ref_pos_dict[chrom_stop_pos][abba] += 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
162 else:
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
163 ref_pos_dict[chrom_stop_pos][abba] = 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
164 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
165 ref_pos_dict[chrom_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
166 ref_pos_dict[chrom_stop_pos][abba] = 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
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 count_indel += 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
169 print("coverage at pos %s = %s, ref = %s, alt = %s, other = %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
170 (pileupcolumn.pos, count_ref + count_alt, count_ref, count_alt, count_indel))
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
171
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
172 # if mutation is in DCS file but not in SSCS, then set counts to NA
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 if chrom_stop_pos not in mut_pos_dict.keys():
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 mut_pos_dict[chrom_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
175 mut_pos_dict[chrom_stop_pos]["ab"] = 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
176 mut_pos_dict[chrom_stop_pos]["ba"] = 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
177 ref_pos_dict[chrom_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
178 ref_pos_dict[chrom_stop_pos]["ab"] = 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
179 ref_pos_dict[chrom_stop_pos]["ba"] = 0
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
180 bam.close()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
181
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
182 # save counts
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
183 with open(sscs_counts_json, "w") as f:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
184 json.dump((mut_pos_dict, ref_pos_dict), f)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
185
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
186
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
187 if __name__ == '__main__':
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
188 sys.exit(mut2sscs(sys.argv))