def main():
usage = "usage: %prog [options] "
parser = argparse.ArgumentParser(description='Given a gzipped vcf file and pedigree file, generate a new vcf with only those samples present in the pedigree (ped file) ')
parser.add_argument('-ped', dest='pedfile', type=str, help="*.ped file")
parser.add_argument('vcfile', type=str,help='*.vcf.gz file')
args=parser.parse_args()
""" parse the pedfile and return the list of iids to keep from the VCF file """
pedobj=Pedfile(args.pedfile)
pedobj.parsePedfile()
keeplist= pedobj.returnIndivids()
#open the VCFfile
vcfh=gzip.open(args.vcfile,'r')
vcfobj=VcfFile(args.vcfile)
vcfobj.parseMetaAndHeaderLines(vcfh)
samples=vcfobj.getSampleList()
newsamples= [ s for s in samples if s in keeplist]
print newsamples
vcfobj.setSampleList(newsamples)
header=vcfobj.returnHeader()
print header
for vrec in vcfobj.yieldVcfRecordwithGenotypes(vcfh):
keepGenotypes=[]
vrec_ziptuple=vrec.zipGenotypes(samples)
for (s, genObj) in vrec_ziptuple:
if s in keeplist:
keepGenotypes.append( genObj )
vrec.addGenotypeList( keepGenotypes )
print vrec.toStringwithGenotypes()
def main():
""" remove samples from a vcf file """
usage = "usage: %prog [options] file.vcf.gz "
# parser = OptionParser(usage)
parser = argparse.ArgumentParser(description="remove samples from vcf file")
parser.add_argument("removesamples", metavar="sample", type=str, nargs="+", help="sample names to remove")
parser.add_argument("-vcf", dest="vcfile", type=str, help="vcf file to remove samples from")
# parser.add_argument("vcf", help="vcf file to analyze")
args = parser.parse_args()
# print 'remove these samples: ', args.samples
# print args.vcfile
vcfh = gzip.open(args.vcfile, "r")
vcfobj = VcfFile(args.vcfile)
vcfobj.parseMetaAndHeaderLines(vcfh)
# print header
samples = vcfobj.getSampleList()
newsamples = [s for s in samples if s not in args.removesamples]
# print 'keep these samples: ', newsamples
vcfobj.setSampleList(newsamples)
header = vcfobj.returnHeader()
print header
for vrec in vcfobj.yieldVcfRecordwithGenotypes(vcfh):
keepGenotypes = []
vrec_ziptuple = vrec.zipGenotypes(samples)
for (s, genObj) in vrec_ziptuple:
if s not in args.removesamples:
# print s
keepGenotypes.append(genObj)
# print keepGenotypes
vrec.addGenotypeList(keepGenotypes)
print vrec.toStringwithGenotypes()