def create_from_file(self, file=None, fmt='gaf', skim=True, **args):
"""
Creates from a file.
Arguments
---------
file : str or file
input file or filename
format : str
name of format e.g. gaf
"""
p = None
if fmt == 'gaf':
p = GafParser()
elif fmt == 'gpad':
p = GpadParser()
elif fmt == 'hpoa':
p = HpoaParser()
else:
logging.error("Format not recognized: {}".format(fmt))
logging.info("Parsing {} with {}/{}".format(file, fmt, p))
if skim:
results = p.skim(file)
return self.create_from_tuples(results, **args)
else:
assocs = p.parse(file, skipheader=True)
return self.create_from_assocs(assocs, **args)
def create_from_remote_file(self, group, snapshot=True, **args):
"""
Creates from remote GAF
"""
import requests
url = "http://snapshot.geneontology.org/annotations/{}.gaf.gz".format(group)
r = requests.get(url, stream=True, headers={'User-Agent': get_user_agent(modules=[requests], caller_name=__name__)})
p = GafParser()
results = p.skim(r.raw)
return self.create_from_tuples(results, **args)
def test_skim_gaf():
p = GafParser()
p.config.ecomap = EcoMap()
results = p.skim(open(POMBASE, "r"))
assert len(results) == 370
for r in results:
print(str(r))
(s, sn, o) = r
assert o.startswith('GO:')
assert s.startswith('PomBase:')
def test_skim_gaf_qualifiers():
p = GafParser()
p.config.ecomap = EcoMap()
p.config.remove_double_prefixes = True
results = p.skim(open(QGAF, "r"))
for r in results:
print(str(r))
(s, sn, o) = r
assert o.startswith('GO:')
assert s.startswith('MGI:') or s.startswith('PomBase')
assert len(results) == 5 # ensure NOTs are skipped
p.config.exclude_relations = ['contributes_to', 'colocalizes_with']
results = p.skim(open(QGAF, "r"))
for r in results:
(s, sn, o) = r
assert o.startswith('GO:')
assert s.startswith('MGI:') or s.startswith('PomBase')
assert len(results) == 3 # ensure NOTs and excludes relations skipped
def create_from_remote_file(self, group, snapshot=True, **args):
"""
Creates from remote GAF
"""
import requests
url = "http://snapshot.geneontology.org/annotations/{}.gaf.gz".format(group)
r = requests.get(url, stream=True)
p = GafParser()
results = p.skim(r.raw)
return self.create_from_tuples(results, **args)
