Mercurial > repos > mheinzl > variant_analyzer2

annotate mut2read.py @ 5:d9cbf833624e draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
author mheinzl
date Tue, 27 Oct 2020 12:46:55 +0000
parents e5953c54cfb5
children 11a2a34f8a2b
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
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
14 0.2.1 2019-10-27 Gundula Povysil -
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 mut2read.py DCS_Mutations.tabular DCS.bam Aligned_Families.tabular Interesting_Reads.fastq
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
18 tag_count_dict.json
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 import argparse
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
22 import json
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
23 import os
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
24 import sys
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
25
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
26 import numpy as np
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
27 import pysam
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():
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
31 parser = argparse.ArgumentParser(description='Takes a tabular 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.')
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',
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
33 help='TABULAR file with DCS mutations.')
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 mut file
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
65 with open(file1, 'r') as mut:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
66 mut_array = np.genfromtxt(mut, skip_header=1, delimiter='\t', comments='#', dtype=str)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
67
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
68 # read dcs bam file
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
69 # pysam.index(file2)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
70 bam = pysam.AlignmentFile(file2, "rb")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
71
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
72 # get tags
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
73 tag_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
74 cvrg_dict = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
75
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
76 if mut_array.shape == (1,13):
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
77 mut_array = mut_array.reshape((1, len(mut_array)))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
78
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
79 for m in range(len(mut_array[:, 0])):
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
80 print(str(m + 1) + " of " + str(len(mut_array[:, 0])))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
81 chrom = mut_array[m, 1]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
82 stop_pos = mut_array[m, 2].astype(int)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
83 chrom_stop_pos = str(chrom) + "#" + str(stop_pos)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
84 ref = mut_array[m, 9]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
85 alt = mut_array[m, 10]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
86
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
87 dcs_len = []
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
88
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
89 for pileupcolumn in bam.pileup(chrom.tostring(), stop_pos - 2, stop_pos, max_depth=100000000):
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
90
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
91 if pileupcolumn.reference_pos == stop_pos - 1:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
92 count_alt = 0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
93 count_ref = 0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
94 count_indel = 0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
95 count_n = 0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
96 count_other = 0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
97 count_lowq = 0
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
98 print("unfiltered reads=", pileupcolumn.n, "filtered reads=", len(pileupcolumn.pileups),
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
99 "difference= ", len(pileupcolumn.pileups) - pileupcolumn.n)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
100 for pileupread in pileupcolumn.pileups:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
101 if not pileupread.is_del and not pileupread.is_refskip:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
102 # query position is None if is_del or is_refskip is set.
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
103 nuc = pileupread.alignment.query_sequence[pileupread.query_position]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
104 dcs_len.append(len(pileupread.alignment.query_sequence))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
105 if nuc == alt:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
106 count_alt += 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
107 tag = pileupread.alignment.query_name
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
108 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
109 tag_dict[tag][chrom_stop_pos] = alt
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
110 else:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
111 tag_dict[tag] = {}
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
112 tag_dict[tag][chrom_stop_pos] = alt
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
113 elif nuc == ref:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
114 count_ref += 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
115 elif nuc == "N":
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
116 count_n += 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
117 elif nuc == "lowQ":
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
118 count_lowq += 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
119 else:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
120 count_other += 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
121 else:
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
122 count_indel += 1
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
123 dcs_median = np.median(np.array(dcs_len))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
124 cvrg_dict[chrom_stop_pos] = (count_ref, count_alt, dcs_median)
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
125
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
126 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" %
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
127 (pileupcolumn.pos, count_ref + count_alt, count_ref, count_alt, count_other, count_n,
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
128 count_indel, count_lowq, dcs_median))
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
129 bam.close()
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
130
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
131 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
132 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
133
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
134 # 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
135 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
136 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
137 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
138 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
139 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
140 tag = splits[0]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
141
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
142 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
143 str1 = splits[4]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
144 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
145 str2 = splits[5]
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
146 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
147
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
148 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
149 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
150 out.write("+" + "\n")
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
151 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
152
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
153
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
154 if __name__ == '__main__':
e5953c54cfb5 planemo upload for repository https://github.com/gpovysil/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
mheinzl
parents:
diff changeset
155 sys.exit(mut2read(sys.argv))