def test_cas_to_chebi():
assert chebi_client.get_chebi_id_from_cas('23261-20-3') == '18035'
assert chebi_client.get_chebi_id_from_cas('100-51-6') == '17987'
assert chebi_client.get_chebi_id_from_cas('-1') is None
def test_cas_to_chebi():
assert chebi_client.get_chebi_id_from_cas('23261-20-3') == '18035'
assert chebi_client.get_chebi_id_from_cas('100-51-6') == '17987'
assert chebi_client.get_chebi_id_from_cas('-1') is None
def _urn_to_db_refs(urn):
"""Converts a Medscan URN to an INDRA db_refs dictionary with grounding
information.
Parameters
----------
url : str
A Medscan URN
Returns
-------
db_refs : dict
A dictionary with grounding information, mapping databases to database
identifiers. If the Medscan URN is not recognized, returns an empty
dictionary.
db_name : str
The Famplex name, if available; otherwise the HGNC name if available;
otherwise None
"""
# Convert a urn to a db_refs dictionary
if urn is None:
return {}, None
p = 'urn:([^:]+):([^:]+)'
m = re.match(p, urn)
if m is None:
return None, None
urn_type = m.group(1)
urn_id = m.group(2)
db_refs = {}
db_name = None
# TODO: support more types of URNs
if urn_type == 'agi-cas':
# Identifier is CAS, convert to CHEBI
chebi_id = get_chebi_id_from_cas(urn_id)
if chebi_id:
db_refs['CHEBI'] = 'CHEBI:%s' % chebi_id
elif urn_type == 'agi-llid':
# This is an Entrez ID, convert to HGNC
hgnc_id = get_hgnc_from_entrez(urn_id)
if hgnc_id is not None:
db_refs['HGNC'] = hgnc_id
# Convert the HGNC ID to a Uniprot ID
uniprot_id = get_uniprot_id(hgnc_id)
db_refs['UP'] = uniprot_id
# Try to lookup HGNC name; if it's available, set it to the
# agent name
db_name = get_hgnc_name(hgnc_id)
elif urn_type == 'agi-ncimorgan':
# Identifier is MESH
db_refs['MESH'] = urn_id
elif urn_type == 'agi-ncimcelltype':
# Identifier is MESH
db_refs['MESH'] = urn_id
elif urn_type == 'agi-meshdis':
# Identifier is MESH
db_refs['MESHDIS'] = urn_id
elif urn_type == 'agi-gocomplex':
# Identifier is GO
db_refs['GO'] = 'GO:%s' % urn_id
elif urn_type == 'agi-go':
# Identifier is GO
db_refs['GO'] = 'GO:%s' % urn_id
elif urn_type == 'agi-ncimtissue':
# Identifier is MESH
db_refs['MESH'] = urn_id
# If we have a GO or MESH grounding, see if there is a corresponding
# Famplex grounding
db_sometimes_maps_to_famplex = ['GO', 'MESH']
for db in db_sometimes_maps_to_famplex:
if db in db_refs:
key = (db, db_refs[db])
if key in famplex_map:
db_refs['FPLX'] = famplex_map[key]
# If the urn corresponds to an eccode, groudn to famplex if that eccode
# is in the Famplex equivalences table
if urn.startswith('urn:agi-enz'):
tokens = urn.split(':')
eccode = tokens[2]
key = ('ECCODE', eccode)
if key in famplex_map:
db_refs['FPLX'] = famplex_map[key]
# If the Medscan URN itself maps to a Famplex id, add a Famplex grounding
key = ('MEDSCAN', urn)
if key in famplex_map:
db_refs['FPLX'] = famplex_map[key]
# If there is a Famplex grounding, use Famplex for entity name
if 'FPLX' in db_refs:
db_name = db_refs['FPLX']
return db_refs, db_name
def _urn_to_db_refs(urn):
"""Converts a Medscan URN to an INDRA db_refs dictionary with grounding
information.
Parameters
----------
urn : str
A Medscan URN
Returns
-------
db_refs : dict
A dictionary with grounding information, mapping databases to database
identifiers. If the Medscan URN is not recognized, returns an empty
dictionary.
db_name : str
The Famplex name, if available; otherwise the HGNC name if available;
otherwise None
"""
# Convert a urn to a db_refs dictionary
if urn is None:
return {}, None
m = URN_PATT.match(urn)
if m is None:
return None, None
urn_type, urn_id = m.groups()
db_refs = {}
db_name = None
# TODO: support more types of URNs
if urn_type == 'agi-cas':
# Identifier is CAS, convert to CHEBI
chebi_id = get_chebi_id_from_cas(urn_id)
if chebi_id:
db_refs['CHEBI'] = 'CHEBI:%s' % chebi_id
db_name = get_chebi_name_from_id(chebi_id)
elif urn_type == 'agi-llid':
# This is an Entrez ID, convert to HGNC
hgnc_id = get_hgnc_from_entrez(urn_id)
if hgnc_id is not None:
db_refs['HGNC'] = hgnc_id
# Convert the HGNC ID to a Uniprot ID
uniprot_id = get_uniprot_id(hgnc_id)
if uniprot_id is not None:
db_refs['UP'] = uniprot_id
# Try to lookup HGNC name; if it's available, set it to the
# agent name
db_name = get_hgnc_name(hgnc_id)
elif urn_type in ['agi-meshdis', 'agi-ncimorgan', 'agi-ncimtissue',
'agi-ncimcelltype']:
if urn_id.startswith('C') and urn_id[1:].isdigit():
# Identifier is probably UMLS
db_refs['UMLS'] = urn_id
else:
# Identifier is MESH
urn_mesh_name = unquote(urn_id)
mesh_id, mesh_name = mesh_client.get_mesh_id_name(urn_mesh_name)
if mesh_id:
db_refs['MESH'] = mesh_id
db_name = mesh_name
else:
db_name = urn_mesh_name
elif urn_type == 'agi-gocomplex':
# Identifier is GO
db_refs['GO'] = 'GO:%s' % urn_id
elif urn_type == 'agi-go':
# Identifier is GO
db_refs['GO'] = 'GO:%s' % urn_id
# If we have a GO or MESH grounding, see if there is a corresponding
# Famplex grounding
db_sometimes_maps_to_famplex = ['GO', 'MESH']
for db in db_sometimes_maps_to_famplex:
if db in db_refs:
key = (db, db_refs[db])
if key in famplex_map:
db_refs['FPLX'] = famplex_map[key]
# If the urn corresponds to an eccode, groudn to famplex if that eccode
# is in the Famplex equivalences table
if urn.startswith('urn:agi-enz'):
tokens = urn.split(':')
eccode = tokens[2]
key = ('ECCODE', eccode)
if key in famplex_map:
db_refs['FPLX'] = famplex_map[key]
# If the Medscan URN itself maps to a Famplex id, add a Famplex grounding
key = ('MEDSCAN', urn)
if key in famplex_map:
db_refs['FPLX'] = famplex_map[key]
# If there is a Famplex grounding, use Famplex for entity name
if 'FPLX' in db_refs:
db_name = db_refs['FPLX']
elif 'GO' in db_refs:
db_name = go_client.get_go_label(db_refs['GO'])
return db_refs, db_name
