model_validation/codescripts/bed_for_bamReadcount.py

import sys,getopt
import os
import re
import fileinput

def usage():
    print(
        """
Usage: python bed_for_bamReadcount.py -i input_vcf_file -o prefix

This script selects SNPs and Indels supported by all callsets.

Please notice that bam-readcount only takes in 1-based coordinates.

Input:
-i a vcf file

Output:
-o  a indel bed file for bam-readcount
        """)

# select supported small variants
def process(oneLine):
    m = re.match('^\#',oneLine)
    if m is not None:
        pass
    else:
        line = oneLine.rstrip()
        strings = line.strip().split('\t')
        # convert the position to bed file for bam-readcount
        # deletion
        if len(strings[3]) > 1 and len(strings[4]) == 1:
            pos = int(strings[1]) + 1
            outline = strings[0] + '\t' + str(pos) + '\t' + str(pos) + '\t' + strings[3] + '\t' + strings[4]+'\n'
            outINDEL.write(outline)
        # insertion
        elif len(strings[3]) == 1 and len(strings[4]) > 1 and (',' not in strings[4]):
            outline = strings[0] + '\t' + strings[1] + '\t' + strings[1] +  '\t' + strings[3] +  '\t' + strings[4] + '\n'
            outINDEL.write(outline)
        else:
            outMNP.write(oneLine)

opts,args = getopt.getopt(sys.argv[1:],"hi:o:")
for op,value in opts:
    if op == "-i":
        inputFile=value
    elif op == "-o":
        prefix=value    
    elif op == "-h":
        usage()
        sys.exit()

if len(sys.argv[1:]) < 3:
    usage()
    sys.exit()

INDELname = prefix + '.bed'
MNPname = prefix + '_MNP.txt'

outINDEL = open(INDELname,'w')
outMNP = open(MNPname,'w')

for line in fileinput.input(inputFile):
    process(line)

outINDEL.close()
outMNP.close()