def get_terms():
terms = generate_mesh_terms(ignore_mappings=True) + \
generate_hgnc_terms() + generate_famplex_terms() + \
generate_uniprot_terms(download=False) + generate_chebi_terms() + \
generate_go_terms()
terms = filter_out_duplicates(terms)
return terms
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)
def test_filter_priority():
term1 = Term('mekk2', 'MEKK2', 'HGNC', '6854', 'MAP3K2', 'previous',
'hgnc', '9606')
term2 = Term('mekk2', 'MEKK2', 'HGNC', '6854', 'MAP3K2', 'synonym', 'up',
'9606')
terms = filter_out_duplicates([term1, term2])
assert len(terms) == 1
term = terms[0]
assert term.status == 'synonym'
key = (me.id, 'FPLX', fplx_entries[0].id)
predicted_mappings[key] = (me, fplx_entries[0])
elif len(hgnc_entries) == 1:
key = (me.id, 'HGNC', hgnc_entries[0].id)
predicted_mappings[key] = (me, hgnc_entries[0])
return predicted_mappings
def find_ambiguities(terms):
ambig_entries = {}
for term in terms:
# We consider it an ambiguity if the same text entry appears
# multiple times
key = term.text
if key in ambig_entries:
ambig_entries[key].append(term)
else:
ambig_entries[key] = [term]
# It's only an ambiguity if there are two entries at least
ambig_entries = {k: v for k, v in ambig_entries.items() if len(v) >= 2}
return ambig_entries
if __name__ == '__main__':
terms = generate_mesh_terms(ignore_mappings=True) + \
generate_hgnc_terms() + generate_famplex_terms() + \
generate_uniprot_terms(download=False)
terms = filter_out_duplicates(terms)
ambigs = find_ambiguities(terms)
mappings = get_mesh_mappings(ambigs)
dump_mappings(mappings, os.path.join(resources, 'mesh_mappings.tsv'))
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)