def parse_results(content):
"""Get terms from a single MTI output"""
if not content:
return []
terms = set()
# Split content into non-empty lines
lines = content.splitlines()
for line in lines:
topic = line.split('|')[1]
# Look for mesh ID of the topic
mesh_id_str = mesh_client.get_mesh_id_name(topic)[0]
if not mesh_id_str:
logger.warning('Mesh ID not found for "%s"' % topic)
continue
# Add mesh ID as a number without prefix
assert mesh_id_str[0] in ['C', 'D'], \
f"Supposedly impossible mesh ID found: {mesh_id_str}"
is_concept = (mesh_id_str[0] == 'C')
mesh_id = int(mesh_id_str[1:])
terms.add((mesh_id, is_concept))
return list(terms)
def get_grounding_from_name(name):
"""Return grounding given an agent name."""
# See if it's a gene name
hgnc_id = get_hgnc_id(name)
if hgnc_id:
return ('HGNC', hgnc_id)
# Check if it's in the grounding map
try:
refs = gm[name]
if isinstance(refs, dict):
for dbn, dbi in refs.items():
if dbn != 'TEXT':
return (dbn, dbi)
# If not, search by text
except KeyError:
pass
chebi_id = get_chebi_id_from_name(name)
if chebi_id:
return ('CHEBI', f'CHEBI:{chebi_id}')
mesh_id, _ = get_mesh_id_name(name)
if mesh_id:
return ('MESH', mesh_id)
return None
def test_mesh_term_name_norm():
# For this one, the corresponding descriptor is D016922, which is in the
# INDRA resource file; however, the descriptor name is "Cellular
# Senescence". This test verifies the expected behavior that in
# offline-only mode, "Cellular Senescence" will return the correct
# descriptor ID, but "Cell Aging" will not, unless using the REST service.
query_name = 'Cellular Senescence'
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_name, offline=True)
assert mesh_id == 'D016922'
assert mesh_name == query_name
query_name = 'Cell Aging'
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_name, offline=True)
assert mesh_id is None
assert mesh_name is None
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_name, offline=False)
assert mesh_id == 'D016922'
assert mesh_name == 'Cellular Senescence'
def test_mesh_term_name_norm():
# For this one, the corresponding descriptor is D016922, which is in the
# INDRA resource file; however, the descriptor name is "Cellular
# Senescence". This test verifies the expected behavior that in
# offline-only mode, "Cellular Senescence" will return the correct
# descriptor ID, but "Cell Aging" will not, unless using the REST service.
query_name = 'Cellular Senescence'
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_name, offline=True)
assert mesh_id == 'D016922'
assert mesh_name == query_name
query_name = 'Cell Aging'
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_name, offline=True)
assert mesh_id is None
assert mesh_name is None
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_name, offline=False)
assert mesh_id == 'D016922'
assert mesh_name == 'Cellular Senescence'
def test_mesh_term_lookups():
queries = {'Breast Cancer': ('D001943', 'Breast Neoplasms'),
'Neoplasms': ('D009369', 'Neoplasms'),
'Colorectal Cancer': ('D015179', 'Colorectal Neoplasms'),
'Intestinal Neoplasms': ('D007414', 'Intestinal Neoplasms'),
'Carcinoma, Non-Small-Cell Lung':
('D002289', 'Carcinoma, Non-Small-Cell Lung'),
'Prostate Cancer': ('D011471', 'Prostatic Neoplasms')}
for query_term, (correct_id, correct_name) in queries.items():
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_term)
assert mesh_id == correct_id
assert mesh_name == correct_name
def test_mesh_term_lookups():
queries = {'Breast Cancer': ('D001943', 'Breast Neoplasms'),
'Neoplasms': ('D009369', 'Neoplasms'),
'Colorectal Cancer': ('D015179', 'Colorectal Neoplasms'),
'Intestinal Neoplasms': ('D007414', 'Intestinal Neoplasms'),
'Carcinoma, Non-Small-Cell Lung':
('D002289', 'Carcinoma, Non-Small-Cell Lung'),
'Prostate Cancer': ('D011471', 'Prostatic Neoplasms')}
for query_term, (correct_id, correct_name) in queries.items():
mesh_id, mesh_name = mesh_client.get_mesh_id_name(query_term)
assert mesh_id == correct_id
assert mesh_name == correct_name
def get_db_refs_by_name(ns, name, node_data):
"""Return standard name and grounding based on a namespace and a name.
Parameters
----------
ns : str
A name space in which the given name is interpreted.
name : str
The name in the given name space to get grounding for.
node_data : dict
Node data for logging purposes.
Returns
-------
name : str
The standardized name for the given entity.
db_refs : dict
The grounding for the given entity.
"""
db_refs = None
if ns == 'HGNC':
# Assumption: name is an HGNC symbol
hgnc_id = hgnc_client.get_current_hgnc_id(name)
if not hgnc_id:
logger.info("Invalid HGNC name: %s (%s)" % (name, node_data))
return name, None
elif isinstance(hgnc_id, list):
logger.info('More than one current HGNC ID for %s, choosing %s' %
(name, hgnc_id[0]))
hgnc_id = hgnc_id[0]
name = hgnc_client.get_hgnc_name(hgnc_id)
db_refs = {'HGNC': hgnc_id}
up_id = _get_up_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
mirbase_id = mirbase_client.get_mirbase_id_from_hgnc_id(hgnc_id)
if mirbase_id:
db_refs['MIRBASE'] = mirbase_id
elif ns in ('UNIPROT', 'UP'):
up_id = None
# This is a simple test to see if name is a valid UniProt ID,
# if we can't get a mnemonic, we assume it's not a UP ID
if uniprot_client.get_mnemonic(name, web_fallback=False):
up_id = name
# We next check if it's a mnemonic
else:
up_id_from_mnem = uniprot_client.get_id_from_mnemonic(name)
if up_id_from_mnem:
up_id = up_id_from_mnem
if not up_id:
logger.info('Couldn\'t get UP ID from %s' % name)
return name, None
db_refs = {'UP': up_id}
hgnc_id = uniprot_client.get_hgnc_id(up_id)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
name = hgnc_client.get_hgnc_name(hgnc_id)
else:
name = uniprot_client.get_gene_name(up_id)
elif ns == 'FPLX':
db_refs = {'FPLX': name}
elif ns in ('GO', 'GOBP', 'GOCC'):
if name == 'cell proliferation':
name = 'cell population proliferation'
go_id = go_client.get_go_id_from_label(name)
if not go_id:
logger.info('Could not find GO ID for %s' % name)
return name, None
db_refs = {'GO': go_id}
name = go_client.get_go_label(go_id)
elif ns in ('MESHPP', 'MESHD', 'MESH'):
mesh_id, mesh_name = mesh_client.get_mesh_id_name(name)
if not mesh_id:
logger.info('Could not find MESH ID from %s' % name)
return name, None
name = mesh_name
db_refs = {'MESH': mesh_id}
# For now, handle MGI/RGD but putting the name into the db_refs so
# it's clear what namespace the name belongs to
# FIXME: Full implementation would look up MGI/RGD identifiers from
# the names, and obtain corresponding Uniprot IDs
elif ns == 'MGI':
up_id = mouse_lookup.get(name)
if up_id:
db_refs = {'UP': up_id}
elif ns == 'RGD':
up_id = rat_lookup.get(name)
if up_id:
db_refs = {'UP': up_id}
# Map Selventa families and complexes to FamPlex
elif ns == 'SFAM':
db_refs = {'SFAM': name}
indra_name = bel_to_indra.get(name)
if indra_name is None:
logger.info('Could not find mapping for BEL/SFAM family: '
'%s (%s)' % (name, node_data))
else:
db_refs['FPLX'] = indra_name
name = indra_name
elif ns == 'SCOMP':
db_refs = {'SCOMP': name}
indra_name = bel_to_indra.get(name)
if indra_name is None:
logger.info('Could not find mapping for BEL/SCOMP complex: '
'%s (%s)' % (name, node_data))
else:
db_refs['FPLX'] = indra_name
name = indra_name
# Map Entrez genes to HGNC/UP
elif ns in ('EGID', 'ENTREZ', 'NCBIGENE'):
hgnc_id = hgnc_client.get_hgnc_from_entrez(name)
db_refs = {'EGID': name}
if hgnc_id is not None:
db_refs['HGNC'] = hgnc_id
name = hgnc_client.get_hgnc_name(hgnc_id)
up_id = hgnc_client.get_uniprot_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
else:
logger.info(
'HGNC entity %s with HGNC ID %s has no '
'corresponding Uniprot ID.', name, hgnc_id)
mirbase_id = mirbase_client.get_mirbase_id_from_hgnc_id(hgnc_id)
if mirbase_id:
db_refs['MIRBASE'] = mirbase_id
else:
logger.debug('Could not map EGID%s to HGNC.' % name)
name = 'E%s' % name
elif ns == 'MIRBASE':
mirbase_id = mirbase_client.get_mirbase_id_from_mirbase_name(name)
if not mirbase_id:
logger.info('Could not map miRBase name %s to ID', name)
return name, None
db_refs = {'MIRBASE': mirbase_id}
hgnc_id = mirbase_client.get_hgnc_id_from_mirbase_id(mirbase_id)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
name = hgnc_client.get_hgnc_name(hgnc_id)
# CHEBI
elif ns == 'CHEBI':
# We first look up BEL's own namespace map for ChEBI names to IDs
chebi_id = chebi_name_id.get(name)
# If that fails, we look up INDRA's ChEBI name to ID mapping
if not chebi_id:
chebi_id = chebi_client.get_chebi_id_from_name(name)
if chebi_id:
db_refs = {'CHEBI': chebi_id}
else:
logger.info('CHEBI name %s not found in map.' % name)
# These appear in the name slot but are actually IDs
elif ns == 'CHEBIID':
chebi_id = identifiers.ensure_chebi_prefix(name)
db_refs = {'CHEBI': chebi_id}
name = chebi_client.get_chebi_name_from_id(chebi_id)
# SDIS, SCHEM: Include the name as the ID for the namespace
elif ns in ('SDIS', 'SCHEM', 'TEXT'):
db_refs = {ns: name}
elif ns == 'TAX':
tid = taxonomy_client.get_taxonomy_id(name)
if tid:
db_refs = {'TAXONOMY': tid}
else:
logger.info('Could not get taxonomy ID for %s' % name)
else:
logger.info("Unhandled namespace: %s: %s (%s)" % (ns, name, node_data))
return name, db_refs
def test_mesh_term_local_missing():
mesh_term = 'XXXX' # dummy term to make sure we don't have it offline
mesh_id, mesh_name = mesh_client.get_mesh_id_name(mesh_term, offline=True)
assert mesh_id is None
assert mesh_name is None
def test_mesh_term_lookup_local():
mesh_term = 'Glucosylceramides'
(mesh_id, mesh_name) = mesh_client.get_mesh_id_name(mesh_term,
offline=True)
assert mesh_id == 'D005963'
assert mesh_name == mesh_term
def get_db_refs_by_name(ns, name, node_data):
"""Return standard name and grounding based on a namespace and a name.
Parameters
----------
ns : str
A name space in which the given name is interpreted.
name : str
The name in the given name space to get grounding for.
node_data : dict
Node data for logging purposes.
Returns
-------
name : str
The standardized name for the given entity.
db_refs : dict
The grounding for the given entity.
"""
db_refs = None
if ns == 'HGNC':
hgnc_id = hgnc_client.get_hgnc_id(name)
if not hgnc_id:
logger.info("Invalid HGNC name: %s (%s)" % (name, node_data))
return name, None
db_refs = {'HGNC': hgnc_id}
up_id = _get_up_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
mirbase_id = mirbase_client.get_mirbase_id_from_hgnc_id(hgnc_id)
if mirbase_id:
db_refs['MIRBASE'] = mirbase_id
elif ns in ('UNIPROT', 'UP'):
up_id = None
gene_name = uniprot_client.get_gene_name(name)
if gene_name:
up_id = name
else:
up_id_from_mnem = uniprot_client.get_id_from_mnemonic(name)
if up_id_from_mnem:
up_id = up_id_from_mnem
gene_name = uniprot_client.get_gene_name(up_id)
if not up_id:
logger.info('Couldn\'t get UP ID from %s' % name)
return name, None
db_refs = {'UP': up_id}
if uniprot_client.is_human(up_id):
hgnc_id = hgnc_client.get_hgnc_id(gene_name)
if not hgnc_id:
logger.info('Uniprot ID linked to invalid human gene '
'name %s' % name)
else:
db_refs['HGNC'] = hgnc_id
elif ns == 'FPLX':
db_refs = {'FPLX': name}
elif ns in ('GO', 'GOBP', 'GOCC'):
go_id = go_client.get_go_id_from_label(name)
if not go_id:
logger.info('Could not find GO ID for %s' % name)
return name, None
db_refs = {'GO': go_id}
elif ns in ('MESHPP', 'MESHD', 'MESH'):
mesh_id = mesh_client.get_mesh_id_name(name)
if not mesh_id:
logger.info('Could not find MESH ID fro %s' % name)
return name, None
db_refs = {'MESH': mesh_id}
# For now, handle MGI/RGD but putting the name into the db_refs so
# it's clear what namespace the name belongs to
# FIXME: Full implementation would look up MGI/RGD identifiers from
# the names, and obtain corresponding Uniprot IDs
elif ns in ('MGI', 'RGD'):
db_refs = {ns: name}
# Map Selventa families to FamPlexes
elif ns == 'SFAM':
db_refs = {'SFAM': name}
indra_name = bel_to_indra.get(name)
if indra_name is None:
logger.info('Could not find mapping for BEL/SFAM family: '
'%s (%s)' % (name, node_data))
else:
db_refs['FPLX'] = indra_name
name = indra_name
# Map Entrez genes to HGNC/UP
elif ns in ('EGID', 'ENTREZ', 'NCBIGENE'):
hgnc_id = hgnc_client.get_hgnc_from_entrez(name)
db_refs = {'EGID': name}
if hgnc_id is not None:
db_refs['HGNC'] = hgnc_id
name = hgnc_client.get_hgnc_name(hgnc_id)
up_id = hgnc_client.get_uniprot_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
else:
logger.info('HGNC entity %s with HGNC ID %s has no '
'corresponding Uniprot ID.',
name, hgnc_id)
mirbase_id = mirbase_client.get_mirbase_id_from_hgnc_id(hgnc_id)
if mirbase_id:
db_refs['MIRBASE'] = mirbase_id
else:
logger.info('Could not map EGID%s to HGNC.' % name)
name = 'E%s' % name
elif ns == 'MIRBASE':
mirbase_id = mirbase_client.get_mirbase_id_from_mirbase_name(name)
if not mirbase_id:
logger.info('Could not map miRBase name %s to ID', name)
return
db_refs = {'MIRBASE': mirbase_id}
hgnc_id = mirbase_client.get_hgnc_id_from_mirbase_id(mirbase_id)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
# CHEBI
elif ns == 'CHEBI':
chebi_id = chebi_name_id.get(name)
if not chebi_id:
chebi_id = chebi_client.get_chebi_id_from_name(name)
if chebi_id:
db_refs = {'CHEBI': chebi_id}
else:
logger.info('CHEBI name %s not found in map.' % name)
# SDIS, SCHEM: Include the name as the ID for the namespace
elif ns in ('SDIS', 'SCHEM'):
db_refs = {ns: name}
else:
logger.info("Unhandled namespace: %s: %s (%s)" % (ns, name,
node_data))
return name, db_refs
def test_mesh_term_local_missing():
mesh_term = 'XXXX' # dummy term to make sure we don't have it offline
mesh_id, mesh_name = mesh_client.get_mesh_id_name(mesh_term, offline=True)
assert mesh_id is None
assert mesh_name is None
def test_mesh_term_lookup_local():
mesh_term = 'Glucosylceramides'
(mesh_id, mesh_name) = mesh_client.get_mesh_id_name(mesh_term, offline=True)
assert mesh_id == 'D005963'
assert mesh_name == mesh_term
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_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
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
def test_mesh_term_local_missing():
mesh_term = 'Rectal Tumors'
mesh_id, mesh_name = mesh_client.get_mesh_id_name(mesh_term, offline=True)
assert mesh_id is None
assert mesh_name is None
def test_mesh_term_local_missing():
mesh_term = 'Rectal Tumors'
mesh_id, mesh_name = mesh_client.get_mesh_id_name(mesh_term, offline=True)
assert mesh_id is None
assert mesh_name is None
