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
""".split())
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():
#!/bin/env python27
import sys,traceback
from getwiki import GlycoMotifWiki, GlyGenMotif
w = GlycoMotifWiki()
from dataset import TSVFileTable
current = set()
for fn in sys.argv[1:]:
rows = TSVFileTable(fn)
for r in rows:
accession = "%06d"%(int(r['accession']),)
prefname = r['name'].strip()
name = [prefname]
gtc = r['gtc'].strip()
redend = r['redend'].strip()
aglycon = r['aglycon'].strip()
for h in r.keys():
if h.startswith('altname'):
if r[h].strip():
name.append(r[h].strip())
motif = GlyGenMotif(accession=accession,prefname=prefname,name=name,glytoucan=gtc,redend=redend,aglycon=aglycon)
if w.update(motif):
print >>sys.stderr, accession
current.add(accession)
for m in w.itermotif(collection=GlyGenMotif):
if m.get('accession') not in current:
print >>sys.stderr, "Deleting:",m.get('pagename')
w.delete(m.get('pagename'))
#!/bin/env python27
import sys,traceback
from getwiki import GlycoMotifWiki, GlyGenMotif
w = GlycoMotifWiki()
from dataset import TSVFileTable
rows = TSVFileTable(sys.argv[1])
current = set()
for r in rows:
accession = "%06d"%(int(r['accession']),)
name = [r['name'].strip()]
gtc = r['gtc'].strip()
redend = r['redend'].strip()
aglycon = r['aglycon'].strip()
for h in r.keys():
if h.startswith('altname'):
name.append(r[h].strip())
motif = GlyGenMotif(accession=accession,name=name,glytoucan=gtc,redend=redend,aglycon=aglycon)
if w.update(motif):
print >>sys.stderr, accession
current.add(accession)
for m in w.itermotif(collection=GlyGenMotif):
if m.get('accession') not in current:
print >>sys.stderr, "Deleting:",m.get('pagename')
w.delete(m.get('pagename'))
