Mercurial > repos > mheinzl > variant_analyzer2

annotate mut2sscs.py @ 65:712a37137b1f draft

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
author mheinzl
date Thu, 18 Mar 2021 10:21:12 +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 """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
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
26 import sys
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
27
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
28 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
29 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
30 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
31
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 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
34 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
35 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
36 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
37 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
38 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
39 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
40 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
41 return parser
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
42
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 def mut2sscs(argv):
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
45 parser = make_argparser()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
46 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
47
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
48 file1 = args.mutFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
49 file2 = args.bamFile
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
50 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
51
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
52 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
53 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
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(file2) 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(file2))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
57
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
58 # 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
59 # 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
60 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
61
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
62 # get tags
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
63 mut_pos_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
64 ref_pos_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
65
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
66 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
67 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
68 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
69 #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
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]
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) + "#" + ref + "#" + alt
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
76
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
77 if len(ref) == len(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
78
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
79 for pileupcolumn in bam.pileup(chrom, stop_pos - 1, stop_pos + 1, max_depth=1000000000):
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
80 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
81 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
82 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
83 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
84 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
85 "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
86 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
87 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
88 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
89 abba = tag[-2:]
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
90 # 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
91 if pileupread.alignment.query_sequence[pileupread.query_position] == alt:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
92 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
93 if chrom_stop_pos in mut_pos_dict:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
94 if abba in mut_pos_dict[chrom_stop_pos]:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
95 mut_pos_dict[chrom_stop_pos][abba] += 1
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
96 else:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
97 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
98 else:
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
99 mut_pos_dict[chrom_stop_pos] = {}
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
100 mut_pos_dict[chrom_stop_pos][abba] = 1
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
101 if chrom_stop_pos not in ref_pos_dict:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
102 ref_pos_dict[chrom_stop_pos] = {}
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
103 ref_pos_dict[chrom_stop_pos][abba] = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
104
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
105 elif pileupread.alignment.query_sequence[pileupread.query_position] == ref:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
106 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
107 if chrom_stop_pos in ref_pos_dict:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
108 if abba in ref_pos_dict[chrom_stop_pos]:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
109 ref_pos_dict[chrom_stop_pos][abba] += 1
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
110 else:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
111 ref_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
112 else:
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
113 ref_pos_dict[chrom_stop_pos] = {}
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
114 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
115 else:
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
116 count_indel += 1
0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
117
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
118 print("coverage at pos %s = %s, ref = %s, alt = %s, indel = %s,\n" %
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
119 (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
120
6
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
121 # if mutation is in DCS file but not in SSCS, then set counts to NA
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
122 if chrom_stop_pos not in mut_pos_dict.keys():
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
123 mut_pos_dict[chrom_stop_pos] = {}
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
124 mut_pos_dict[chrom_stop_pos]["ab"] = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
125 mut_pos_dict[chrom_stop_pos]["ba"] = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
126 ref_pos_dict[chrom_stop_pos] = {}
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
127 ref_pos_dict[chrom_stop_pos]["ab"] = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
128 ref_pos_dict[chrom_stop_pos]["ba"] = 0
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
129 else:
11a2a34f8a2b planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents: 0
diff changeset
130 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
131 bam.close()
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 # save counts
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
134 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
135 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
136
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
137
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
138 if __name__ == '__main__':
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
139 sys.exit(mut2sscs(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
140