edit genotype in gvcf

addallgt.py

#!/usr/bin/env python

## get individual genotype from raw vcf
## output is tab-delimited chr \t start \t end \t ref \t alt \t GT \t ref,alt

usage="""
read in tab-delimited bedfile: chr,start,end,ref,alt,gt
usage: python addallgt.py <infile> <outfile>
"""
import sys
import os

#if os.path.isfile(sys.argv[2]):
	#exit('output file already exists')

if len(sys.argv) < 2:
        print usage


infile = open(sys.argv[1],'r')	
outfile = open(sys.argv[2],'w')


linecount = 0
multalleles=0

for line in infile:
	lineitems = line.split()
	cleanedLine = line.replace("\n","")

	ref = lineitems[3]
	alt = lineitems[4]
	

	if not (len(lineitems[3]) == 1):
		continue

	if lineitems[5]=="0/0":
		outfile.write(cleanedLine + "\t" + ref + "," + ref + "\n")

	elif lineitems[5]=="0/1":
		outfile.write(cleanedLine + "\t" + ref + "," + alt + "\n")
			
	elif lineitems[5]=="1/0":
		outfile.write(cleanedLine + "\t" + ref + "," + alt + "\n")

	elif lineitems[5]=="1/1":
		outfile.write(cleanedLine + "\t" + alt + "," + alt + "\n")
		
	else:
		#print "%s warning, invalid item at %s " % (lineitems[5], linecount)
		multalleles += 1
	
	linecount += 1
	
print "%s" % (multalleles)


infile.close()
outfile.close()