def test_get_hgnc_id():
hgnc_id = uniprot_client.get_hgnc_id('P07305')
assert hgnc_id == '4714', hgnc_id
# NRXN2: ['P58401', 'Q9P2S2'] 8009
hgnc_id = uniprot_client.get_hgnc_id('P58401')
assert hgnc_id == '8009', hgnc_id
hgnc_id = uniprot_client.get_hgnc_id('Q9P2S2')
assert hgnc_id == '8009', hgnc_id
def iter_terms(force: bool = False) -> Iterable[Term]:
"""Iterate over NCI PID terms."""
hgnc_id_to_name = get_id_name_mapping("hgnc")
hgnc_name_to_id = {v: k for k, v in hgnc_id_to_name.items()}
for uuid, cx in iter_networks(force=force, use_tqdm=True):
name = None
for node in iterate_aspect(cx, "networkAttributes"):
if node["n"] == "name":
name = node["v"]
term = Term(reference=Reference(prefix=PREFIX,
identifier=uuid,
name=name), )
genes = set()
for node in iterate_aspect(cx, "nodes"):
name, reference = node["n"], node["r"]
hgnc_id = hgnc_name_to_id.get(name)
if hgnc_id:
genes.add((hgnc_id, name))
elif any(reference.startswith(x) for x in ("CHEBI:", "cas:")):
pass
elif reference.startswith("uniprot:"):
uniprot_id = reference[len("uniprot:"):]
hgnc_id = get_hgnc_id(uniprot_id)
if hgnc_id is None: # this only happens for proteins that seem to be virus related
# TODO reinvestigate this later
logger.debug(
"uniprot could not map %s/%s/%s to HGNC",
name,
reference,
get_gene_name(uniprot_id, web_fallback=False),
)
else:
name = hgnc_id_to_name[hgnc_id]
genes.add((hgnc_id, name))
else:
logger.debug(f"unmapped: {name}, {reference}")
for hgnc_id, hgnc_symbol in genes:
term.append_relationship(has_part,
Reference("hgnc", hgnc_id, hgnc_symbol))
yield term
def iter_terms() -> Iterable[Term]:
"""Iterate over NCI PID terms."""
hgnc_id_to_name = get_id_name_mapping('hgnc')
hgnc_name_to_id = {v: k for k, v in hgnc_id_to_name.items()}
for uuid, cx in iter_networks(use_tqdm=True):
name = None
for node in iterate_aspect(cx, 'networkAttributes'):
if node['n'] == 'name':
name = node['v']
term = Term(reference=Reference(prefix=PREFIX,
identifier=uuid,
name=name), )
genes = set()
for node in iterate_aspect(cx, 'nodes'):
name, reference = node['n'], node['r']
hgnc_id = hgnc_name_to_id.get(name)
if hgnc_id:
genes.add((hgnc_id, name))
elif any(reference.startswith(x) for x in ('CHEBI:', 'cas:')):
pass
elif reference.startswith('uniprot:'):
uniprot_id = reference[len('uniprot:'):]
hgnc_id = get_hgnc_id(uniprot_id)
if hgnc_id is None: # this only happens for proteins that seem to be virus related
# TODO reinvestigate this later
logger.debug(
'uniprot could not map %s/%s/%s to HGNC',
name,
reference,
get_gene_name(uniprot_id, web_fallback=False),
)
else:
name = hgnc_id_to_name[hgnc_id]
genes.add((hgnc_id, name))
else:
logger.debug(f'unmapped: {name}, {reference}')
for hgnc_id, hgnc_symbol in genes:
term.append_relationship(pathway_has_part,
Reference('hgnc', hgnc_id, hgnc_symbol))
yield term