def fix_id(db_ns: str, db_id: str) -> Tuple[str, str]:
"""Fix ID issues specific to the SIF dump."""
if db_ns == "GO":
if db_id.isnumeric():
db_id = "0" * (7 - len(db_id)) + db_id
elif db_ns == "EFO" and db_id.startswith("EFO:"):
db_id = db_id[4:]
elif db_ns == "UP" and db_id.startswith("SL"):
db_ns = "UPLOC"
elif db_ns == "UP" and "-" in db_id and not db_id.startswith("SL-"):
db_id = db_id.split("-")[0]
elif db_ns == 'FPLX' and db_id == 'TCF-LEF':
db_id = 'TCF_LEF'
db_id = ensure_prefix_if_needed(db_ns, db_id)
return db_ns, db_id
def process_selventa_xref(xref):
if pandas.isna(xref):
return ''
db_refs = {}
for xref_part in xref.split('|'):
prefix, db_id = xref_part.split(':', maxsplit=1)
ns = xref_mappings.get(prefix)
if not ns:
logger.info('Unknown namespace: %s' % prefix)
continue
db_id = ensure_prefix_if_needed(ns, db_id)
db_refs[ns] = db_id
assert_valid_db_refs(db_refs)
db_refs_str = '|'.join(['%s:%s' % (k, v)
for k, v in sorted(db_refs.items())])
return db_refs_str
def fix_invalidities_db_refs(db_refs: Mapping[str, str]) -> Mapping[str, str]:
"""Return a fixed version of a db_refs grounding dict."""
if 'PUBCHEM' in db_refs and \
db_refs['PUBCHEM'].startswith('CID'):
db_refs['PUBCHEM'] = \
db_refs['PUBCHEM'].replace('CID:', '').strip()
db_refs = {k: v for k, v in db_refs.items()
if v is not None}
for k, v in copy.deepcopy(db_refs).items():
if k == 'CHEMBL' and not v.startswith('CHEMBL'):
db_refs[k] = 'CHEMBL%s' % v
elif k == 'ECCODE':
db_refs['ECCODE'] = db_refs['ECCODE'].replace('.-', '')
elif k == 'UNIPROT':
db_refs.pop(k)
# This is really a location
if v.startswith('SL-'):
db_refs['UPLOC'] = v
# Otherwise we just fix the invalid key
else:
db_refs['UP'] = v
elif k == 'UP':
# There are cases where this is an empty string
if not v.strip():
db_refs.pop('UP', None)
# Sometimes we have two IDs separated by a comma
if ',' in v:
db_refs['UP'] = v.split(',')[0]
if v.startswith('SL-'):
db_refs['UPLOC'] = db_refs.pop('UP')
elif k == 'UAZ':
db_refs.pop('UAZ')
if v.startswith('CVCL'):
db_refs['CVCL'] = v
elif k == 'TAXONOMY' and v == '-1':
db_refs.pop('TAXONOMY', None)
elif k == 'LINCS' and re.match(r'\d+-\d+', v):
db_refs['HMS-LINCS'] = db_refs.pop('LINCS')
elif k == 'CVCL' and re.match(r'^[A-Z0-9]{4}$', v):
db_refs['CVCL'] = 'CVCL_%s' % v
elif k == 'CO':
db_refs['CL'] = 'CL:%s' % db_refs.pop('CO')
elif k == 'FPLX' and '-' in v:
db_refs['FPLX'] = v.replace('-', '_')
elif k == 'DRUGBANK' and v.startswith('DBSALT'):
db_refs['DRUGBANK.SALT'] = db_refs.pop('DRUGBANK')
# For MGI and RGD some sources added names as IDs that are invalid
# and not easily fixable without reverse lookups so we rather
# remove these.
elif k == 'MGI' and not re.match(
identifiers_registry['mgi']['pattern'], v):
db_refs.pop('MGI', None)
elif k == 'RGD' and not re.match(
identifiers_registry['rgd']['pattern'], v):
db_refs.pop('RGD', None)
else:
new_val = ensure_prefix_if_needed(k, v)
db_refs[k] = new_val
return db_refs
def fix_invalidities(stmts):
for stmt in stmts:
for agent in stmt.real_agent_list():
for db_ns, db_id in agent.db_refs.items():
agent.db_refs[db_ns] = ensure_prefix_if_needed(db_ns, db_id)
return stmts
def test_ensure_prefix_if_needed():
assert ensure_prefix_if_needed('CHEBI', 'CHEBI:123') == 'CHEBI:123'
assert ensure_prefix_if_needed('CHEBI', '123') == 'CHEBI:123'
assert ensure_prefix_if_needed('GO', '00004') == 'GO:00004'
assert ensure_prefix_if_needed('EFO', '1234') == '1234'
assert ensure_prefix_if_needed('XXXX', '1234') == '1234'