if metafile:
logger.info('Reading metafile ...')
metadata = dict(
TsvReader(metafile, cnames=True, row=lambda r: tuple(
(r.IID, r))).dump())
else:
metadata = None
logger.info('Reading genotype matrix ...')
# snp1 gt1s1 gt1s2 ...
inreader = TsvReader(infile, cnames=True)
samples = inreader.meta[1:]
logger.info('Writing tfam file ...')
tfamWriter = TsvWriter(tfamfile)
tfamWriter.meta = ['FID', 'IID', 'PID', 'MID', 'Sex', 'Pheno']
#tfamWriter.writeHead(callback = lambda meta: '#' + '\t'.join(meta))
if not metadata:
for s in samples:
tfamWriter.write([s, s, '0', '0', 'other', '-9'])
else:
for s in samples:
tfamWriter.write([
metadata[s].FID if s in metadata and 'FID' in metadata[s] else s,
s, (metadata[s].PID or '0')
if s in metadata and 'PID' in metadata[s] else '0',
(metadata[s].MID or '0')
if s in metadata and 'MID' in metadata[s] else '0',
(metadata[s].Sex or 'other')
if s in metadata and 'Sex' in metadata[s] else 'other',
(metadata[s].Pheno or '-9')
indata1 = TsvReader(infile1, **inopts1) indata2 = TsvReader(infile2, **inopts2) cnames1 = indata1.meta if not rnames1 else indata1.meta[1:] cnames2 = indata2.meta if not rnames2 else indata2.meta[1:] paired = list(set(cnames1) & set(cnames2)) cnames1 = cnames2 = paired if rnames1: cnames1 = [indata1.meta[0]] + cnames1 if rnames2: cnames2 = [indata2.meta[0]] + cnames2 cindex1 = [indata1.meta.index(c) for c in cnames1] cindex2 = [indata2.meta.index(c) for c in cnames2] outdata1 = TsvWriter(outfile1) outdata2 = TsvWriter(outfile2) outdata1.meta = cnames1 outdata2.meta = cnames2 outdata1.writeHead() outdata2.writeHead() for r1 in indata1: outdata1.write(r1[i] for i in cindex1) outdata1.close() for r2 in indata2: outdata2.write(r2[i] for i in cindex2) outdata2.close()