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