def ReadTSV(filename):
snvheaders = [_f for _f in """CHROM POS REF ALT""".split() if _f]
base, extn = filename.rsplit('.', 1)
extn = extn.lower()
if extn == 'csv':
snvs = CSVFileTable(filename=filename)
elif extn == 'tsv':
snvs = TSVFileTable(filename=filename)
elif extn == 'xls':
snvs = XLSFileTable(filename=filename)
elif extn == 'xlsx':
snvs = XLSXFileTable(filename=filename)
elif extn == 'txt':
snvs = TXTFileTable(filename=filename, headers=snvheaders)
else:
raise RuntimeError("Unexpected SNV file extension: %s" % filename)
for h in snvheaders:
if h not in snvs.headers():
raise RuntimeError("Required header: %s missing from SNV file %s" %
(h, filename))
assert (snvs.headers()[:4] == snvheaders)
chrom = set()
snvdata = []
for r in snvs:
ri = list(map(r.get, snvs.headers()))
chrom.add(ri[0])
snvdata.append(ri)
return ["\t".join(snvs.headers())], chrom, snvdata
snvdata = {}
# extrasnvheaders = []
# usedsnvheaders = set()
snvchroms = defaultdict(set)
for filename in opt.snvs:
base, extn = filename.rsplit('.', 1)
extn = extn.lower()
if extn == 'csv':
snvs = CSVFileTable(filename=filename)
elif extn == 'vcf':
snvs = VCFFile(filename=filename)
elif extn == 'tsv':
snvs = TSVFileTable(filename=filename)
elif extn == 'xls':
snvs = XLSFileTable(filename=filename)
elif extn == 'xlsx':
snvs = XLSXFileTable(filename=filename)
elif extn == 'txt':
snvs = TXTFileTable(filename=filename, headers=snvheaders)
else:
raise RuntimeError("Unexpected SNV file extension: %s" % filename)
for h in snvheaders:
if h not in snvs.headers():
raise RuntimeError("Required header: %s missing from SNV file %s" %
(h, filename))
for h in snvs.headers():
if h in snvheaders:
continue