Mercurial > repos > in_silico > cravat_vcf_convert

comparison cravat_convert/cravat_convert.py @ 20:a0b39b857c7a draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author in_silico
date Wed, 18 Jul 2018 10:17:38 -0400
parents
children
comparison
equal deleted inserted replaced
19:5de40a6dd491 20:a0b39b857c7a
1 '''
2 Convert a VCF format file to Cravat format file
3 '''
4
5 import os
6 import argparse
7 from vcf_converter import CravatConverter
8 from __future__ import print_function
9
10 def get_vcf_mapping():
11 """ : VCF Headers mapped to their index position in a row of VCF values.
12 : These are only the mandatory columns, per the VCF spec.
13 """
14 return {
15 'CHROM': 0,
16 'POS': 1,
17 'ID': 2,
18 'REF': 3,
19 'ALT': 4,
20 'QUAL': 5,
21 'FILTER': 6,
22 'INFO': 7
23 }
24
25
26 def get_args():
27 parser = argparse.ArgumentParser()
28 parser.add_argument('--input',
29 '-i',
30 required = True,
31 help='Input path to a VCF file for conversion',)
32 parser.add_argument('--output',
33 '-o',
34 default = None,
35 help = 'Output path to write the cravat file to')
36 return parser.parse_args()
37
38
39 def convert(in_path, out_path=None, cr_sep='\t', cr_newline='\n'):
40 """ : Convert a VCF file to a Cravat file.
41 : Arguments:
42 : in_path: <str> path to input vcf file
43 : out_path: <str> path to output cravat file. Will defualt to cravat_converted.txt in the input directory.
44 : cr_sep: <str> the value delimiter for the output cravat file. Default value of '\\t'.
45 : out_newline: <str> the newline delimiter in the output cravat file. Default of '\\n'
46 """
47 if not out_path:
48 base, _ = os.path.split(in_path)
49 out_path = os.path.join(base, "cravat_converted.txt")
50
51 with open(in_path, 'r') as in_file, \
52 open(out_path, 'w') as out_file:
53
54 # cr_count will be used to generate the 'TR' field of the cravat rows (first header)
55 cr_count = 0
56 # VCF lines are always assumed to be '+' strand, as VCF doesn't specify that attribute
57 strand = '+'
58 # VCF converter. Adjusts position, reference, and alternate for Cravat formatting.
59 converter = CravatConverter()
60 # A dictionary of mandatory vcf headers mapped to their row indices
61 vcf_mapping = get_vcf_mapping()
62
63 for line in in_file:
64 if line.startswith("#"):
65 continue
66 line = line.strip().split()
67 # row is dict of VCF headers mapped to corresponding values of this line
68 row = { header: line[index] for header, index in vcf_mapping.items() }
69 for alt in row["ALT"].split(","):
70 new_pos, new_ref, new_alt = converter.extract_vcf_variant(strand, row["POS"], row["REF"], alt)
71 new_pos, new_ref, new_alt = str(new_pos), str(new_ref), str(new_alt)
72 cr_line = cr_sep.join([
73 'TR' + str(cr_count), row['CHROM'], new_pos, strand, new_ref, new_alt, row['ID']
74 ])
75 out_file.write(cr_line + cr_newline)
76 cr_count += 1
77
78
79 if __name__ == "__main__":
80 cli_args = get_args()
81 if cli_args.output == None:
82 base, _ = os.path.split(cli_args.input)
83 cli_args.output = os.path.join(base, "cravat_converted.txt")
84 convert(in_path = cli_args.input, out_path = cli_args.output)