def _propagate_go_term_ancestors(self, mapping):
"""Given a mapping object which maps entities to GO terms, this
method ensures that all the ancestor terms of each GO term
appear at the respective entities. Returns a copy of the mapping
which contains these modifications."""
result = bidict(mapping)
for entity, terms in result.iteritems_left():
ancestors = self.tree.ancestors(*terms)
result.add_left_multi(entity, ancestors)
return result
def _propagate_go_term_ancestors(self, mapping):
"""Given a mapping object which maps entities to GO terms, this
method ensures that all the ancestor terms of each GO term
appear at the respective entities. Returns a copy of the mapping
which contains these modifications."""
result = bidict(mapping)
for entity, terms in result.iteritems_left():
ancestors = self.tree.ancestors(*terms)
result.add_left_multi(entity, ancestors)
return result
def read_goa_file(self, goa_file, ev_codes):
"""Reads the GOA file, return a defaultdict that,
for each protein id it has a set of strings with the
associated GO terms
"""
d = bidict()
for line in open_anything(goa_file):
if not line.startswith(("!", "#")):
# split line, obtain protein_id, go_term, and ev_code
fields = line.split("\t", 7)
prot_id, goterm, evcode = fields[1], fields[4], fields[6]
if evcode in ev_codes:
d.add_left(prot_id, self.go_tree.lookup(goterm))
self.log.info("GOA file read. " + str(d.len_left()) + " proteins loaded")
return d
def read_goa_file(self, goa_file, ev_codes):
"""Reads the GOA file, return a defaultdict that,
for each protein id it has a set of strings with the
associated GO terms
"""
goa_mapping = bidict()
for line in open_anything(goa_file):
if not line.startswith(("!", "#")):
# split line, obtain protein_id, go_term, and ev_code
fields = line.split("\t", 7)
prot_id, goterm, evcode = fields[1], fields[4], fields[6]
if evcode in ev_codes:
goa_mapping.add_left(prot_id, self.go_tree.lookup(goterm))
self.log.info("GOA file read. %s proteins loaded",
str(goa_mapping.len_left()))
return goa_mapping
