def main(infile, outfile, annotations, descriptions, categories):
cats = dict()
with open(categories, 'rU') as handle:
for line in handle:
if line.startswith('#') or line.strip() == '':
continue
line = line.rstrip('\n')
record = cat_line(line)
cats[record['catid']] = record
annots = dict()
with open(annotations, 'rU') as handle:
for line in handle:
if line.startswith('#') or line.strip() == '':
continue
line = line.rstrip('\n')
record = annot_line(line)
cat = record['catid']
record.update(cats[cat])
annots[record['id']] = record
descs = dict()
with open(descriptions, 'rU') as handle:
for line in handle:
if line.startswith('#') or line.strip() == '':
continue
line = line.rstrip('\n')
record = desc_line(line)
record['category'] = ''
record['short_catname'] = ''
record['long_catname'] = ''
descs[record['id']] = record
ips = InterproscanResult(infile)
with outhandler(outfile) as handle:
for query, analyses in ips.items():
results = list()
if 'SUPERFAMILY' in analyses:
analysis = analyses['SUPERFAMILY']
sfids = set()
for record in analysis:
acc = record.accession.lstrip('SSF')
sfids.add(acc)
for acc in sfids:
try:
results.append(annots[acc])
except KeyError:
results.append(descs[acc])
template = '{seqid}\t{category}\t{short_catname}\t{long_catname}\t{id}\t{name}\n'
for result in results:
handle.write(template.format(seqid=query, **result))
return
def main(infile, outfile, pantherfile):
pantherdb = External2GO(pantherfile, fmt='panther')
ips = InterproscanResult(infile)
with outhandler(outfile) as handle:
for query, analyses in ips.items():
if 'PANTHER' in analyses:
panther_ids = [f.accession for f in analyses['PANTHER']]
panther_names = [pantherdb[f].name for f in panther_ids]
unnamed = [
i for i, val in enumerate(panther_names) if
val in {'FAMILY NOT NAMED', 'SUBFAMILY NOT NAMED'}
]
subfamily = [i for i, val in enumerate(panther_ids) if ":" in val]
family = [i for i, val in enumerate(panther_ids) if ":" not in val]
represented_fams = set()
ids = list()
names = list()
for sf in subfamily:
fam = panther_ids[sf].split(':')[0]
if sf in unnamed or panther_ids[sf] in represented_fams:
continue
represented_fams.add(fam)
represented_fams.add(panther_ids[sf])
ids.append(panther_ids[sf])
names.append(panther_names[sf])
for f in family:
if f in unnamed or panther_ids[f] in represented_fams:
continue
represented_fams.add(panther_ids[f])
ids.append(panther_ids[f])
names.append(panther_names[f])
else:
names = []
ids = []
if len(names) == 0 and len(ids) == 0:
continue
template = '{seqid}\t{ids}\t{names}\n'
for id_, name in zip(ids, names):
handle.write(template.format(
seqid=query,
ids=id_,
names=name,
))
def main(
infile,
outfile,
obofile,
outfmt='long',
pantherfile=None,
pfamfile=None,
smartfile=None,
interprofile=None,
prositefile=None,
printsfile=None,
prodomfile=None,
tigrfamfile=None,
pirsffile=None,
hamapfile=None,
domainfile=None,
datadir=None,
):
""" . """
if datadir is None:
datadir = ''
dbs = dict()
if pantherfile is not None:
pantherdb = External2GO(pjoin(datadir, pantherfile), fmt='panther')
dbs['PANTHER'] = pantherdb
if pfamfile is not None:
pfamdb = External2GO(pjoin(datadir, pfamfile))
dbs['Pfam'] = pfamdb
if smartfile is not None:
smartdb = External2GO(pjoin(datadir, smartfile))
dbs['SMART'] = smartdb
if interprofile is not None:
interprodb = External2GO(pjoin(datadir, interprofile))
dbs['IPR'] = interprodb
if prositefile is not None:
prositedb = External2GO(pjoin(datadir, prositefile))
dbs['ProSitePatterns'] = prositedb
dbs['ProSiteProfiles'] = prositedb
if printsfile is not None:
printsdb = External2GO(pjoin(datadir, printsfile))
dbs['PRINTS'] = printsdb
if prodomfile is not None:
prodomdb = External2GO(pjoin(datadir, prodomfile))
dbs['ProDom'] = prodomdb
if tigrfamfile is not None:
tigrfamdb = External2GO(pjoin(datadir, tigrfamfile))
dbs['TIGRFAM'] = tigrfamdb
if pirsffile is not None:
pirsfdb = External2GO(pjoin(datadir, pirsffile))
dbs['PIRSF'] = pirsfdb
if hamapfile is not None:
hamapdb = External2GO(pjoin(datadir, hamapfile))
dbs['Hamap'] = hamapdb
if domainfile is not None:
domaindb = External2GO(pjoin(datadir, domainfile), fmt='superfamily')
dbs['SUPERFAMILY'] = domaindb
ips = InterproscanResult(infile)
godag = GODag(pjoin(datadir, obofile))
with outhandler(outfile) as handle:
for query, analyses in ips.items():
ontologies = set()
for analysis, records in analyses.items():
for record in records:
if analysis not in dbs:
continue
acc = record.accession
if acc not in dbs[analysis]:
continue
gos = [g.id for g in dbs[analysis][acc].ontologies]
for go in gos:
domain = godag[go].namespace.replace('_', ' ')
term = godag[go].name
ontologies.add((go, term, domain))
if len(ontologies) == 0:
continue
if outfmt == 'long':
template = "{seqid}\t{go}\t{term}\t{domain}\n"
for ontology in ontologies:
go, term, domain = ontology
handle.write(template.format(
seqid=query,
go=go,
term=term,
domain=domain,
))
elif outfmt == 'association':
template = "{seqid}\t{gos}\n"
gos = [go for go, term, domain in ontologies]
handle.write(template.format(
seqid=query,
gos=';'.join(gos)
))
