def get_specific_chebi_id(chebi_ids, name):
# NOTE: this function is mainly factored out to be able to use cacheing, it
# requires a frozenset as input to work.
# First, if we have a manual override, we just do that
manual_id = manual_chebi_map.get(name)
if manual_id:
return manual_id
# The first thing we do is eliminate the secondary IDs by mapping them to
# primaries
primary_ids = {chebi_client.get_primary_id(cid) for cid in chebi_ids}
# Occasinally, invalid ChEBI IDs are given that don't have corresponding
# primary IDs, which we can filter out
primary_ids = {pi for pi in primary_ids if pi is not None}
# We then get rid of generic IDs which are never useful for grounding
non_generic_ids = primary_ids - generic_chebi_ids
# We then try name-based grounding to see if any of the names in the list
# match the name of the entity well enough
grounding_names = [
chebi_client.get_chebi_name_from_id(p) for p in non_generic_ids
]
for grounding_name, grounding_id in zip(grounding_names, non_generic_ids):
if grounding_name and (name.lower() == grounding_name.lower()):
return grounding_id
# If we still have no best grounding, we try to distill the IDs down to
# the most specific one based on the hierarchy
specific_chebi_id = chebi_client.get_specific_id(non_generic_ids)
return specific_chebi_id
def test_specific_chebi_ids():
ids = ['76971', '37045', '15996', '75771', '37121', '57600']
spec_id = chebi_client.get_specific_id(ids)
assert spec_id == 'CHEBI:15996', spec_id