|
12
|
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
|
|
|
9 # File read/write configuration variables
|
|
|
10 vcf_sep = '\t'
|
|
|
11 cr_sep = '\t'
|
|
|
12 cr_newline = '\n'
|
|
|
13
|
|
|
14 # VCF Headers mapped to their index position in a row of VCF values
|
|
|
15 vcf_mapping = {
|
|
|
16 'CHROM': 0,
|
|
|
17 'POS': 1,
|
|
|
18 'ID': 2,
|
|
|
19 'REF': 3,
|
|
|
20 'ALT': 4,
|
|
|
21 'QUAL': 5,
|
|
|
22 'FILTER': 6,
|
|
|
23 'INFO': 7,
|
|
|
24 'FORMAT': 8,
|
|
|
25 'NA00001': 9,
|
|
|
26 'NA00002': 10,
|
|
|
27 'NA00003': 11
|
|
|
28 }
|
|
|
29
|
|
|
30
|
|
|
31 def get_args():
|
|
|
32 parser = argparse.ArgumentParser()
|
|
|
33 parser.add_argument('--input',
|
|
|
34 '-i',
|
|
|
35 required = True,
|
|
|
36 help='Input path to a VCF file for conversion',)
|
|
|
37 parser.add_argument('--output',
|
|
|
38 '-o',
|
|
|
39 default = os.path.join(os.getcwd(), "cravat_converted.txt"),
|
|
|
40 help = 'Output path to write the cravat file to')
|
|
|
41 return parser.parse_args()
|
|
|
42
|
|
|
43
|
|
|
44 def convert(in_path, out_path=None):
|
|
|
45 if not out_path:
|
|
|
46 base, _ = os.path.split(in_path)
|
|
|
47 out_path = os.path.join(base, "cravat_converted.txt")
|
|
|
48
|
|
|
49 with open(in_path, 'r') as in_file, \
|
|
|
50 open(out_path, 'w') as out_file:
|
|
|
51
|
|
|
52 # cr_count will be used to generate the 'TR' field of the cravat rows (first header)
|
|
|
53 cr_count = 0
|
|
|
54 # VCF lines are always assumed to be '+' strand, as VCF doesn't specify that attribute
|
|
|
55 strand = '+'
|
|
|
56 # VCF converter. Adjusts position, reference, and alternate for Cravat formatting.
|
|
|
57 converter = CravatConverter()
|
|
|
58
|
|
|
59 for line in in_file:
|
|
|
60 if line.startswith("#"):
|
|
|
61 continue
|
|
|
62 line = line.strip().split(vcf_sep)
|
|
|
63 # row is dict of VCF headers mapped to corresponding values of this line
|
|
|
64 row = { header: line[index] for header, index in vcf_mapping.items() }
|
|
|
65 for alt in row["ALT"].split(","):
|
|
|
66 new_pos, new_ref, new_alt = converter.extract_vcf_variant(strand, row["POS"], row["REF"], alt)
|
|
|
67 new_pos, new_ref, new_alt = str(new_pos), str(new_ref), str(new_alt)
|
|
|
68 cr_line = cr_sep.join([
|
|
|
69 'TR' + str(cr_count), row['CHROM'], new_pos, strand, new_ref, new_alt, row['ID']
|
|
|
70 ])
|
|
|
71 out_file.write(cr_line + cr_newline)
|
|
|
72 cr_count += 1
|
|
|
73
|
|
|
74
|
|
|
75 if __name__ == "__main__":
|
|
|
76 cli_args = get_args()
|
|
|
77 convert(cli_args.input, cli_args.output)
|
