haploid2diploid.py

import sys
import argparse
import re
import subprocess as sp


def main(args):
    IN = sys.stdin if args.vcf=="-" else sp.Popen("bcftools view %(vcf)s" % vars(args),shell=True,stdout=sp.PIPE).stdout
    for l in IN:
        try:
            l = l.decode()
        except:
            pass
        row = l.strip().split()
        if l[0]=="#":
            sys.stdout.write(l)
            continue
        gt_index = row[8].split(":").index("GT")
        for i in range(9,len(row)):
            tmp = row[i].split(":")
            gt = "%s/%s" % (tmp[gt_index],tmp[gt_index])
            tmp[gt_index] = gt
            row[i] = ":".join(tmp)
        sys.stdout.write("\t".join(row)+"\n")



parser = argparse.ArgumentParser(description='Changed haploid to diploid calls',formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('vcf',help='VCF file (use \'-\' if piping from stdin)')
parser.set_defaults(func=main)
args = parser.parse_args()
args.func(args)