def _get_agent_from_ref(self, ref):
# TODO: handle collections
if ref.attrib.get('category') == 'collection':
logger.warning('Skipping collection Agent.')
return None
name_tag = ref.find("var/[@name='name']")
if name_tag is not None:
name = name_tag.text
else:
return None
uid_tag = ref.find("var/[@name='uid']")
if uid_tag is not None:
uid = uid_tag.text
else:
uid = None
db_refs = {}
text_tag = ref.find("var/[@name='raw-text']")
if text_tag is not None:
db_refs['TEXT'] = text_tag.text
if uid is not None and uid.startswith('UP:'):
up_mnemonic = uid[3:]
up_id = uniprot_client.get_id_from_mnemonic(up_mnemonic)
if up_id is not None:
up_name = uniprot_client.get_gene_name(up_id)
if up_name is not None:
name = up_name
db_refs['UP'] = up_id
assert name is not None
agent = Agent(name, db_refs=db_refs)
return agent
def _get_agent_from_ref(self, ref):
# TODO: handle collections
if ref.attrib.get('category') == 'collection':
logger.warning('Skipping collection Agent.')
return None
name_tag = ref.find("var/[@name='name']")
if name_tag is not None:
name = name_tag.text
else:
return None
uid_tag = ref.find("var/[@name='uid']")
if uid_tag is not None:
uid = uid_tag.text
else:
uid = None
db_refs = {}
text_tag = ref.find("var/[@name='raw-text']")
if text_tag is not None:
db_refs['TEXT'] = text_tag.text
if uid is not None and uid.startswith('UP:'):
up_mnemonic = uid[3:]
up_id = uniprot_client.get_id_from_mnemonic(up_mnemonic)
if up_id is not None:
up_name = uniprot_client.get_gene_name(up_id)
if up_name is not None:
name = up_name
db_refs['UP'] = up_id
assert name is not None
agent = Agent(name, db_refs=db_refs)
return agent
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 _get_agent(self, participant):
dbid = participant.get('identifier')
text = participant.get('entity_text')[0]
if dbid == 'GENERIC':
if not text:
return None
else:
return Agent(text)
db_refs = {}
entity_type = participant.get('entity_type')
if entity_type in ['protein', 'chemical', 'gene']:
# TODO: standardize name here
name = participant.get('entity_text')[0]
db_refs['TEXT'] = text
if dbid:
db_name, db_id = dbid.split(':')
if db_name.lower() == 'uniprot':
uniprot_id = uniprot_client.get_id_from_mnemonic(db_id)
db_refs['UP'] = uniprot_id
elif db_name.lower() == 'pubchem':
chebi_id = chebi_client.get_chebi_id_from_pubchem(db_id)
db_refs['CHEBI'] = chebi_id
elif db_name.lower() == 'hgnc':
db_refs['HGNC'] = db_id
elif entity_type == 'protein_family':
name = text
else:
return None
# TODO: handle other participant types
agent = Agent(name, db_refs=db_refs)
features = participant.get('features')
if features:
for feature in features:
feature_type = feature.get('feature_type')
if feature_type == 'modification_feature':
mc = self._get_mod_condition(feature)
agent.mods.append(mc)
elif feature_type == 'binding_feature':
bc = self._get_bound_condition(feature)
agent.bound_conditions.append(bc)
elif feature_type == 'mutation_feature':
mc = self._get_mut_condition(feature)
agent.mutations.append(mc)
elif feature_type == 'location_feature':
agent.location = feature.get('location')
not_features = participant.get('features')
if not_features:
for feature in not_features:
feature_type = feature.get('feature_type')
if feature_type == 'modification_feature':
mc = self._get_mod_condition(feature)
mc.is_modified = False
agent.mods.append(mc)
elif feature_type == 'binding_feature':
bc = self._get_bound_condition(feature)
bc.is_bound = False
agent.bound_conditions.append(bc)
return agent
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 _get_agent(self, participant):
dbid = participant.get('identifier')
text = participant.get('entity_text')[0]
if dbid == 'GENERIC':
if not text:
return None
else:
return Agent(text)
db_refs = {}
entity_type = participant.get('entity_type')
if entity_type in ['protein', 'chemical', 'gene']:
# TODO: standardize name here
name = participant.get('entity_text')[0]
db_refs['TEXT'] = text
if dbid:
db_name, db_id = dbid.split(':')
if db_name.lower() == 'uniprot':
uniprot_id = uniprot_client.get_id_from_mnemonic(db_id)
db_refs['UP'] = uniprot_id
elif db_name.lower() == 'pubchem':
chebi_id = chebi_client.get_chebi_id_from_pubchem(db_id)
db_refs['CHEBI'] = chebi_id
elif db_name.lower() == 'hgnc':
db_refs['HGNC'] = db_id
elif entity_type == 'protein_family':
name = text
else:
return None
# TODO: handle other participant types
agent = Agent(name, db_refs=db_refs)
features = participant.get('features')
if features:
for feature in features:
feature_type = feature.get('feature_type')
if feature_type == 'modification_feature':
mc = self._get_mod_condition(feature)
agent.mods.append(mc)
elif feature_type == 'binding_feature':
bc = self._get_bound_condition(feature)
agent.bound_conditions.append(bc)
elif feature_type == 'mutation_feature':
mc = self._get_mut_condition(feature)
agent.mutations.append(mc)
elif feature_type == 'location_feature':
agent.location = feature.get('location')
not_features = participant.get('features')
if not_features:
for feature in not_features:
feature_type = feature.get('feature_type')
if feature_type == 'modification_feature':
mc = self._get_mod_condition(feature)
mc.is_modified = False
agent.mods.append(mc)
elif feature_type == 'binding_feature':
bc = self._get_bound_condition(feature)
bc.is_bound = False
agent.bound_conditions.append(bc)
return agent
def _get_agent_from_ref(self, ref):
# TODO: handle collections
if ref.attrib.get('category') == 'collection':
#logger.warning('Skipping collection Agent.')
return None
# Find the name, uid and raw-text tags first and get their text
# content if available
uid_tag = ref.find("var/[@name='uid']")
name_tag = ref.find("var/[@name='name']")
text_tag = ref.find("var/[@name='raw-text']")
if name_tag is not None and name_tag.text:
name = name_tag.text
else:
name = None
if uid_tag is not None and uid_tag.text:
uid = uid_tag.text
else:
uid = None
if text_tag is not None and text_tag.text:
raw_text = text_tag.text
else:
raw_text = None
# TODO: factor this out and reuse fix_agents
db_refs = {}
# Save raw text if available
if raw_text:
db_refs['TEXT'] = raw_text
agent_name = raw_text
# If we have a proper UID then we try to reconstruct an Agent from that
if uid is not None and len(uid.split(':')) == 2:
db_ns, db_id = uid.split(':')
be_id = famplex_map.get((db_ns, db_id))
if be_id:
db_refs[db_ns] = db_id
db_refs['FPLX'] = be_id
agent_name = be_id
elif db_ns in ['UP', 'Uniprot']:
id_from_mnemonic = uniprot_client.get_id_from_mnemonic(db_id)
if id_from_mnemonic:
db_id = id_from_mnemonic
db_refs['UP'] = db_id
hgnc_id = uniprot_client.get_hgnc_id(db_id)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
agent_name = hgnc_client.get_hgnc_name(hgnc_id)
else:
gene_name = uniprot_client.get_gene_name(db_id)
if gene_name:
agent_name = gene_name
elif db_ns == 'NCIT':
db_refs['NCIT'] = db_id
target = ncit_map.get(db_id)
if target:
db_refs[target[0]] = target[1]
if target[0] == 'HGNC':
up_id = hgnc_client.get_uniprot_id(target[1])
agent_name = hgnc_client.get_hgnc_name(target[1])
if up_id:
db_refs['UP'] = up_id
elif target[0] == 'UP':
agent_name = uniprot_client.get_gene_name(target[1])
if agent_name:
hgnc_id = hgnc_client.get_hgnc_id(agent_name)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
elif db_ns == 'FA':
db_refs['NXP'] = 'FA:' + db_id
elif db_ns == 'XFAM':
db_refs['PF'] = db_id.split('.')[0]
elif db_ns == 'CHEBI':
db_refs['CHEBI'] = 'CHEBI:' + db_id
elif db_ns in ['GO', 'MESH', 'FPLX']:
db_refs[db_ns] = db_id
# Handle old BE mappings and add them as FPLX
elif db_ns == 'BE':
db_refs['FPLX'] = db_id
elif db_ns in ['PR', 'CO', 'CVCL', 'EFO', 'ORPHANET']:
db_refs[db_ns] = db_id
else:
logger.warning('Unknown database name space %s' % db_ns)
if not agent_name:
if raw_text is not None:
agent_name = raw_text
else:
return None
assert (agent_name)
agent = Agent(agent_name, db_refs=db_refs)
return agent
