python 将VCF中的FILTER列复制到所有样本#vcf的新创建的GF FORMAT字段

Posted 2021-05-09

#!bin/usr/python
'''
Filter TransferScript
usage:
bcftools view  <filename> | python filter_transfer.py  | bcftools view -O z > <filename.het.polarized.vcf.gz>

Copies the filter column to a FORMAT (gt) field.

'''
import sys

def main():
    format_added = False
    for l in sys.stdin.xreadlines():
        l = l.strip()
        if l.startswith("#CHROM"):
            # Get Sample information and count
            samples = l.strip().split("\t")[9:]
        elif l.startswith("#"):
            # Add Info line for het polarization flag
            if l.startswith("##FORMAT") and format_added == False:
                format_added = True
                l = l + "\n##FORMAT=<ID=GF,Number=1,Type=String,Description=\"Genotype Filter\">"
        else:
            l = l.split("\t")
            FILTER = l[6]
            l[8] = l[8] + ":" + "GF"
            l[9:] = [x + ":" + FILTER for x in l[9:]]
            l = '\t'.join(l)

        sys.stdout.write(l + "\n")



if __name__ == '__main__':
    main()