def get_grounder(
prefix: Union[str, Iterable[str]],
unnamed: Optional[Iterable[str]] = None,
grounder_cls: Optional[Type[Grounder]] = None,
) -> Grounder:
"""Get a Gilda grounder for the given namespace."""
unnamed = set() if unnamed is None else set(unnamed)
if isinstance(prefix, str):
prefix = [prefix]
terms: List[gilda.term.Term] = []
for p in prefix:
try:
p_terms = list(
get_gilda_terms(p, identifiers_are_names=p in unnamed))
except NoBuild:
continue
else:
terms.extend(p_terms)
terms = filter_out_duplicates(terms)
terms_dict = multidict((term.norm_text, term) for term in terms)
if grounder_cls is None:
return Grounder(terms_dict)
else:
return grounder_cls(terms_dict)
target_prefix = 'uniprot.chain'
relation = 'skos:exactMatch'
source = 'https://github.com/indralab/gilda/blob/master/scripts/' \
'generate_uniprot_chain_proonto_mappings.py'
match_type = 'lexical'
rows = []
pro = obonet.read_obo(PROONTO_OBO)
for pro_id, matches in matches_per_id.items():
target_id = matches[0].term.id
target_name = matches[0].term.entry_name
source_name = pro.nodes[pro_id]['name']
row = (source_prefix, pro_id, source_name, relation, target_prefix,
target_id, target_name, match_type, 0.8, source)
rows.append(row)
append_prediction_tuples(rows, deduplicate=True)
if __name__ == '__main__':
# 1. Parse all the UniProt synonyms that are for human
# protein fragments into Gilda Terms
terms = get_uniprot_terms()
# 2. Instantiate a grounder with these terms
terms_dict = defaultdict(list)
for term in terms:
terms_dict[term.norm_text].append(term)
grounder = Grounder(terms_dict)
# 3. Parse all the Protein Ontology synonyms and ground each of them, then
# store the results
matches_per_id = ground_proonto_terms(grounder)
# 4. Dump spreadsheet with non-ambiguous equivalences in BioMappings format
dump_predictions()
def get_grounder(self):
if self.grounder is None:
self.grounder = Grounder()
return self.grounder
from gilda.grounder import Grounder
from . import appreq
gr = Grounder()
def test_grounder():
entries = gr.lookup('kras')
statuses = [e.status for e in entries]
assert 'assertion' in statuses
for entry in entries:
if entry.status == 'assertion':
assert entry.id == '6407', entry
scores = gr.ground('kras')
assert len(scores) == 1, scores
assert appreq(scores[0].score, 0.9845), scores
scores = gr.ground('k-ras')
assert len(scores) == 1, scores
assert appreq(scores[0].score, 0.9936), scores
scores = gr.ground('KRAS')
assert len(scores) == 1, scores
assert appreq(scores[0].score, 1.0), scores
scores = gr.ground('bRaf')
assert len(scores) == 1, scores
assert appreq(scores[0].score, 0.9936), scores
def test_grounder_bug():
# Smoke test to make sure the 'NA' entry in grounding terms doesn't get
def get_grounder(prefix, url: Optional[str] = None) -> Grounder:
"""Get a Gilda grounder for the given namespace."""
terms = list(get_gilda_terms(prefix, url=url))
terms = filter_out_duplicates(terms)
terms = multidict((term.norm_text, term) for term in terms)
return Grounder(terms)