def validate(db_ns, db_id):
"""Validate identifier, accepting HGNC name or ID"""
if db_ns == 'HGNC':
if db_id.isdigit():
return validate_id(db_ns, db_id)
else:
return get_hgnc_id(db_id) is not None
else:
return validate_id(db_ns, db_id)
def id_url(ag):
# Return identifier URLs in a prioritized order
# TODO: we should add handling for UPPRO here, however, that would require
# access to UniProt client resources in the context of the DB REST API
# which could be problematic
for db_name in link_namespace_order:
if db_name in ag.db_refs:
# Handle a special case where a list of IDs is given
if isinstance(ag.db_refs[db_name], list):
db_id = ag.db_refs[db_name][0]
if db_name == 'WM':
db_id = db_id[0]
else:
db_id = ag.db_refs[db_name]
# We can add more name spaces here if there are issues
if db_name in {'CHEBI'}:
db_id = ensure_prefix('CHEBI', db_id)
# Here we validate IDs to make sure we don't surface invalid
# links.
if not validate_id(db_name, db_id):
logger.debug('Invalid grounding encountered: %s:%s' %
(db_name, db_id))
continue
# Finally, we return a valid identifiers.org URL
return get_identifiers_url(db_name, db_id)
def update_mesh_supplementary_names():
"""Update MeSH ID to name mappings for supplementary terms."""
supp_url = ('ftp://nlmpubs.nlm.nih.gov/online/mesh/MESH_FILES/'
'xmlmesh/supp2021.gz')
supp_path = os.path.join(path, 'mesh_supp2021.gz')
if not os.path.exists(supp_path):
logging.info('Download MeSH supplement from %s', supp_url)
urlretrieve(supp_url, supp_path)
logging.info('Done downloading MeSH supplement')
with gzip.open(supp_path) as supp_file:
logging.info('Parsing MeSH supplement')
supp_et = ET.parse(supp_file)
supp_rows = []
reg_number_mappings = []
for record in supp_et.iterfind('SupplementalRecord'):
uid = record.find('SupplementalRecordUI').text
name = record.find('SupplementalRecordName/String').text
mapped_to_terms = record.findall('HeadingMappedToList/HeadingMappedTo/'
'DescriptorReferredTo/DescriptorUI')
mapped_to = ','.join([term.text.replace('*', '')
for term in mapped_to_terms])
term_name_str = _get_term_name_str(record, name)
reg_number_tags = record.findall('ConceptList/Concept/RegistryNumber')
if len(reg_number_tags) == 1:
reg_number = reg_number_tags[0].text
from indra.statements import validate
if validate.validate_id('CAS', reg_number):
reg_number_mappings.append([uid, reg_number])
supp_rows.append((uid, name, term_name_str, mapped_to))
fname = os.path.join(path, 'mesh_supp_id_label_mappings.tsv')
write_unicode_csv(fname, supp_rows, delimiter='\t')
fname = os.path.join(path, 'mesh_cas_mappings.tsv')
write_unicode_csv(fname, reg_number_mappings, delimiter='\t')
def _make_agent(self, hprd_id, refseq_id=None):
if hprd_id is None or hprd_id is nan:
return None
# Get the basic info (HGNC name/symbol, Entrez ID) from the
# ID mappings dataframe
try:
egid = self.id_df.loc[hprd_id].EGID
except KeyError:
logger.info('HPRD ID %s not found in mappings table.' % hprd_id)
return None
if not egid:
logger.info('No Entrez ID for HPRD ID %s' % hprd_id)
return None
# Get the HGNC ID
hgnc_id = hgnc_client.get_hgnc_from_entrez(egid)
# If we couldn't get an HGNC ID for the Entrez ID, this means that
# the Entrez ID has been discontinued or replaced.
if not hgnc_id:
self.no_hgnc_for_egid.update(egid)
return None
# Get the (possibly updated) HGNC Symbol
hgnc_name = hgnc_client.get_hgnc_name(hgnc_id)
assert hgnc_name is not None
# See if we can get a Uniprot ID from the HGNC symbol--if there is
# a RefSeq ID we wil also try to use it to get an isoform specific
# UP ID, but we will have this one to fall back on. But if we can't
# get one here, then we skip the Statement
up_id_from_hgnc = hgnc_client.get_uniprot_id(hgnc_id)
if not up_id_from_hgnc:
self.no_up_for_hgnc.update((egid, hgnc_name, hgnc_id))
return None
# If we have provided the RefSeq ID, it's because we need to make
# sure that we are getting the right isoform-specific ID (for sequence
# positions of PTMs). Here we try to get the Uniprot ID from the
# Refseq->UP mappings in the protmapper.uniprot_client.
if refseq_id is not None:
if not validate_id('REFSEQ_PROT', refseq_id):
if validate_id('NCBIPROTEIN', refseq_id):
refseq_ns = 'NCBIPROTEIN'
else:
refseq_ns = None
else:
refseq_ns = 'REFSEQ_PROT'
if refseq_ns == 'REFSEQ_PROT':
# Get the Uniprot IDs from the uniprot client
up_ids = uniprot_client.get_ids_from_refseq(refseq_id,
reviewed_only=True)
else:
up_ids = []
# Nothing for this RefSeq ID (quite likely because the RefSeq
# ID is obsolete; take the UP ID from HGNC
if len(up_ids) == 0:
self.no_up_for_refseq.update(refseq_id)
up_id = up_id_from_hgnc
# More than one reviewed entry--no thanks, we'll take the one
# from HGNC instead
elif len(up_ids) > 1:
self.many_ups_for_refseq.update(refseq_id)
up_id = up_id_from_hgnc
# We got a unique, reviewed UP entry for the RefSeq ID
else:
up_id = up_ids[0]
# If it's the canonical isoform, strip off the '-1'
if up_id.endswith('-1'):
up_id = up_id.split('-')[0]
# For completeness, get the Refseq ID from the HPRD ID table
else:
refseq_id = self.id_df.loc[hprd_id].REFSEQ_PROTEIN
if not validate_id('REFSEQ_PROT', refseq_id):
if validate_id('NCBIPROTEIN', refseq_id):
refseq_ns = 'NCBIPROTEIN'
else:
refseq_ns = None
else:
refseq_ns = 'REFSEQ_PROT'
up_id = up_id_from_hgnc
# Make db_refs, return Agent
db_refs = {}
if hgnc_id:
db_refs['HGNC'] = hgnc_id
if up_id:
if ',' in up_id:
pass
elif '-' in up_id:
up_base = up_id.split('-')[0]
db_refs['UP'] = up_base
db_refs['UPISO'] = up_id
else:
db_refs['UP'] = up_id
if egid:
db_refs['EGID'] = egid
if refseq_ns and refseq_id:
db_refs[refseq_ns] = refseq_id
return Agent(hgnc_name, db_refs=db_refs)