def _get_db_refs(bpe):
db_refs = {}
if _is_protein(bpe):
hgnc_id = BiopaxProcessor._get_hgnc_id(bpe)
uniprot_id = BiopaxProcessor._get_uniprot_id(bpe)
# Handle missing HGNC/UP ids
if hgnc_id and not uniprot_id:
uniprot_id = hgnc_client.get_uniprot_id(hgnc_id)
if uniprot_id and not hgnc_id:
if uniprot_client.is_human(uniprot_id):
hgnc_name = uniprot_client.get_gene_name(uniprot_id, False)
if hgnc_name:
hgnc_id = hgnc_client.get_hgnc_id(hgnc_name)
if hgnc_id is not None:
db_refs['HGNC'] = hgnc_id
if uniprot_id is not None:
db_refs['UP'] = uniprot_id
elif _is_small_molecule(bpe):
chebi_id = BiopaxProcessor._get_chebi_id(bpe)
if chebi_id is not None:
db_refs['CHEBI'] = chebi_id
else:
chebi_id = BiopaxProcessor._get_chebi_id(bpe)
if chebi_id is not None:
db_refs['CHEBI'] = chebi_id
hgnc_id = BiopaxProcessor._get_hgnc_id(bpe)
if hgnc_id is not None:
db_refs['HGNC'] = hgnc_id
uniprot_id = BiopaxProcessor._get_uniprot_id(bpe)
if uniprot_id is not None:
db_refs['UP'] = uniprot_id
return db_refs
def filter_human_only(stmts_in, **kwargs):
"""Filter out statements that are not grounded to human genes.
Parameters
----------
stmts_in : list[indra.statements.Statement]
A list of statements to filter.
save : Optional[str]
The name of a pickle file to save the results (stmts_out) into.
Returns
-------
stmts_out : list[indra.statements.Statement]
A list of filtered statements.
"""
dump_pkl = kwargs.get('save')
logger.info('Filtering %d statements for human genes only...' %
len(stmts_in))
stmts_out = []
for st in stmts_in:
human_genes = True
for agent in st.agent_list():
if agent is not None:
upid = agent.db_refs.get('UP')
if upid and not uniprot_client.is_human(upid):
human_genes = False
break
if human_genes:
stmts_out.append(st)
logger.info('%d statements after filter...' % len(stmts_out))
if dump_pkl:
dump_statements(stmts_out, dump_pkl)
return stmts_out
def read_phosphosite(fname):
df = pandas.read_csv(fname, index_col=None)
statements = []
antibody_map = {}
for _, row in df.iterrows():
sub_upid = row['SUB_ID']
if not pandas.isnull(sub_upid):
sub_hgnc_symbol = uniprot_client.get_gene_name(sub_upid)
sub_hgnc = hgnc_client.get_hgnc_id(sub_hgnc_symbol)
else:
sub_hgnc_symbol = row['SUB_GENE']
sub_hgnc_id = hgnc_client.get_hgnc_id(sub_hgnc_symbol)
sub_upid = hgnc_client.get_uniprot_id(sub_hgnc_id)
sub = Agent(sub_hgnc_symbol,
db_refs={'UP': sub_upid,'HGNC': sub_hgnc})
residue = row['Actual_site'][0]
if len(row['Actual_site']) > 1:
position = row['Actual_site'][1:]
else:
position = None
sub_readout = deepcopy(sub)
mc = ModCondition('phosphorylation', residue, position)
sub_readout.mods = [mc]
ps = row['phosphosite']
if ps in antibody_map:
found = False
for p in antibody_map[ps]:
if p.name == sub.name and p.mods[0].residue == residue and \
p.mods[0].position == position:
found = True
break
if not found:
antibody_map[ps].append(sub_readout)
else:
antibody_map[ps] = [sub_readout]
kin_upid = row['KIN_ID']
if not pandas.isnull(kin_upid):
if not uniprot_client.is_human(kin_upid):
print('%s non human' % kin_upid)
continue
kin_hgnc_symbol = uniprot_client.get_gene_name(kin_upid)
kin_hgnc = hgnc_client.get_hgnc_id(kin_hgnc_symbol)
else:
kin_hgnc_symbol = row['KINASE_GENE_SYMBOL']
kin_hgnc_id = hgnc_client.get_hgnc_id(kin_hgnc_symbol)
kin_upid = hgnc_client.get_uniprot_id(kin_hgnc_id)
kin = Agent(kin_hgnc_symbol,
db_refs={'UP': kin_upid,'HGNC': kin_hgnc})
ev = Evidence(source_api='phosphosite')
st = Phosphorylation(kin, sub, residue, position, evidence = [ev])
statements.append(st)
return statements, antibody_map
def _human_only_filter(stmts_in):
stmts_out = []
for st in stmts_in:
agents = [a for a in st.agent_list() if a is not None]
non_human = False
for a in agents:
hgnc_id = a.db_refs.get("HGNC")
up_id = a.db_refs.get("UP")
if not hgnc_id:
if up_id and not uniprot_client.is_human(up_id):
non_human = True
break
if not non_human:
stmts_out.append(st)
return stmts_out
def test_all_protein_db_refs():
unmapped_uniprot_ids = []
for obj in bp.model.getObjects().toArray():
bpe = bpc._cast_biopax_element(obj)
if bpc._is_protein(bpe):
db_refs = bpc.BiopaxProcessor._get_db_refs(bpe)
uniprot_id = db_refs.get('UP')
hgnc_id = db_refs.get('HGNC')
if uniprot_id:
if uniprot_client.is_human(uniprot_id):
if not hgnc_id:
unmapped_uniprot_ids.append(uniprot_id)
unmapped_uniprot_ids = sorted(list(set(unmapped_uniprot_ids)))
# The number of unmapped entries should not increase
# so we check for an upper limit here
assert(len(unmapped_uniprot_ids) < 95)
def get_grounding(self):
import indra.databases.hgnc_client as hgc
import indra.databases.uniprot_client as upc
be = self.db_refs.get('FPLX')
if be:
return ('FPLX', be)
hgnc = self.db_refs.get('HGNC')
if hgnc:
if isinstance(hgnc, list):
hgnc = hgnc[0]
return ('HGNC', hgc.get_hgnc_name(str(hgnc)))
up = self.db_refs.get('UP')
if up:
if isinstance(up, list):
up = up[0]
if upc.is_human(up):
gene_name = upc.get_gene_name(up, web_fallback=False)
if gene_name:
return ('HGNC', gene_name)
else:
return ('UP', up)
return (None, None)
def is_non_human_protein(bio_ontology, node):
if bio_ontology.get_ns(node) == 'UP' and \
not uniprot_client.is_human(bio_ontology.get_id(node)):
return True
return False
def test_not_is_human():
assert(not uniprot_client.is_human('P31938'))
def test_noentry_is_human():
assert not uniprot_client.is_human('XXXX')
def test_is_human():
assert uniprot_client.is_human('P00533')
def test_is_human():
assert(uniprot_client.is_human('P00533'))
def test_not_is_human():
assert not uniprot_client.is_human('P31938')
def _get_db_refs(entity_term):
agent_name = entity_term['text']
db_refs = {}
for xr in entity_term['xrefs']:
ns = xr['namespace']
if ns == 'uniprot':
up_id = xr['id']
db_refs['UP'] = up_id
# Look up official names in UniProt
gene_name = up_client.get_gene_name(up_id)
if gene_name is not None:
agent_name = gene_name
# If the gene name corresponds to an HGNC ID, add it to the
# db_refs
if up_client.is_human(up_id):
hgnc_id = hgnc_client.get_hgnc_id(gene_name)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
elif ns == 'hgnc':
hgnc_id = xr['id']
db_refs['HGNC'] = hgnc_id
# Look up the standard gene symbol and set as name
hgnc_name = hgnc_client.get_hgnc_name(hgnc_id)
if hgnc_name:
agent_name = hgnc_name
# Look up the corresponding uniprot id
up_id = hgnc_client.get_uniprot_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
elif ns == 'pfam':
be_id = famplex_map.get(('PF', xr['id']))
if be_id:
db_refs['FPLX'] = be_id
agent_name = be_id
db_refs['PF'] = xr['id']
elif ns == 'interpro':
be_id = famplex_map.get(('IP', xr['id']))
if be_id:
db_refs['FPLX'] = be_id
agent_name = be_id
db_refs['IP'] = xr['id']
elif ns == 'chebi':
db_refs['CHEBI'] = xr['id']
elif ns == 'pubchem':
db_refs['PUBCHEM'] = xr['id']
elif ns == 'go':
db_refs['GO'] = xr['id']
elif ns == 'mesh':
db_refs['MESH'] = xr['id']
elif ns == 'hmdb':
db_refs['HMDB'] = xr['id']
elif ns == 'simple_chemical':
if xr['id'].startswith('HMDB'):
db_refs['HMDB'] = xr['id']
elif ns == 'be':
db_refs['FPLX'] = xr['id']
agent_name = db_refs['FPLX']
# These name spaces are ignored
elif ns in ['uaz']:
pass
else:
logger.warning('Unhandled xref namespace: %s' % ns)
db_refs['TEXT'] = entity_term['text']
return agent_name, db_refs
def _get_agent(node_data, node_modifier_data=None):
# FIXME: Handle translocations on the agent for ActiveForms, turn into
# location conditions
# Check the node type/function
node_func = node_data[pc.FUNCTION]
if node_func not in (pc.PROTEIN, pc.RNA, pc.BIOPROCESS, pc.COMPLEX,
pc.PATHOLOGY, pc.ABUNDANCE, pc.MIRNA):
mod_data = ('No node data' if not node_modifier_data else
node_modifier_data.get(pc.CNAME))
logger.info("Nodes of type %s not handled: %s" % (node_func, mod_data))
return None
# Skip gene/protein fusions
if pc.FUSION in node_data:
logger.info("Gene and protein fusions not handled: %s" %
str(node_data))
return None
# COMPLEXES ------------
# First, handle complexes, which will consist recursively of other agents
if node_func == pc.COMPLEX:
# First, check for members: if there are no members, we assume this
# is a named complex
members = node_data.get(pc.MEMBERS)
if members is None:
return None
# Otherwise, get the "main" agent, to which the other members will be
# attached as bound conditions
main_agent = _get_agent(members[0])
# If we can't get the main agent, return None
if main_agent is None:
return None
bound_conditions = [
BoundCondition(_get_agent(m), True) for m in members[1:]
]
# Check the bound_conditions for any None agents
if any([bc.agent is None for bc in bound_conditions]):
return None
main_agent.bound_conditions = bound_conditions
# Get activity of main agent
ac = _get_activity_condition(node_modifier_data)
main_agent.activity = ac
return main_agent
# OTHER NODE TYPES -----
# Get node identifier information
name = node_data.get(pc.NAME)
ns = node_data[pc.NAMESPACE]
ident = node_data.get(pc.IDENTIFIER)
# No ID present, get identifier using the name, namespace
db_refs = None
if not ident:
assert name, "Node must have a name if lacking an identifier."
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 None
db_refs = {'HGNC': hgnc_id}
up_id = _get_up_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
# FIXME: Look up go ID in ontology lookup service
# FIXME: Look up MESH IDs from name
# FIXME: For now, just use node name
elif ns in ('GOBP', 'MESHPP', 'MESHD'):
db_refs = {}
# 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 == 'EGID':
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))
else:
logger.info('Could not map EGID%s to HGNC.' % name)
name = 'E%s' % name
# CHEBI
elif ns == 'CHEBI':
chebi_id = chebi_name_id.get(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:
print("Unhandled namespace: %s: %s (%s)" % (ns, name, node_data))
# We've already got an identifier, look up other identifiers if necessary
else:
# Get the name, overwriting existing name if necessary
if ns == 'HGNC':
name = hgnc_client.get_hgnc_name(ident)
db_refs = {'HGNC': ident}
up_id = _get_up_id(ident)
if up_id:
db_refs['UP'] = up_id
elif ns == 'UP':
db_refs = {'UP': ident}
name = uniprot_client.get_gene_name(ident)
assert name
if uniprot_client.is_human(ident):
hgnc_id = hgnc_client.get_hgnc_id(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 in ('MGI', 'RGD'):
raise ValueError('Identifiers for MGI and RGD databases are not '
'currently handled: %s' % node_data)
else:
print("Unhandled namespace with identifier: %s: %s (%s)" %
(ns, name, node_data))
if db_refs is None:
logger.info('Unable to get identifier information for node: %s' %
node_data)
return None
# Get modification conditions
mods, muts = _get_all_pmods(node_data)
# Get activity condition
ac = _get_activity_condition(node_modifier_data)
to_loc = _get_translocation_target(node_modifier_data)
# Check for unhandled node modifiers, skip if so
if _has_unhandled_modifiers(node_modifier_data):
return None
# Make the agent
ag = Agent(name,
db_refs=db_refs,
mods=mods,
mutations=muts,
activity=ac,
location=to_loc)
return ag
def test_noentry_is_human():
assert(not uniprot_client.is_human('XXXX'))
def get_agent(node_data, node_modifier_data=None):
# FIXME: Handle translocations on the agent for ActiveForms, turn into
# location conditions
# Check the node type/function
node_func = node_data[pc.FUNCTION]
if node_func not in (pc.PROTEIN, pc.RNA, pc.BIOPROCESS, pc.COMPLEX,
pc.PATHOLOGY, pc.ABUNDANCE, pc.MIRNA):
mod_data = node_modifier_data or 'No node data'
logger.info("Nodes of type %s not handled: %s",
node_func, mod_data)
return None
# Skip gene/protein fusions
if pc.FUSION in node_data:
logger.info("Gene and protein fusions not handled: %s" % str(node_data))
return None
# COMPLEXES ------------
# First, handle complexes, which will consist recursively of other agents
if node_func == pc.COMPLEX:
# First, check for members: if there are no members, we assume this
# is a named complex
members = node_data.get(pc.MEMBERS)
if members is None:
return None
# Otherwise, get the "main" agent, to which the other members will be
# attached as bound conditions
main_agent = get_agent(members[0])
# If we can't get the main agent, return None
if main_agent is None:
return None
bound_conditions = [BoundCondition(get_agent(m), True)
for m in members[1:]]
# Check the bound_conditions for any None agents
if any([bc.agent is None for bc in bound_conditions]):
return None
main_agent.bound_conditions = bound_conditions
# Get activity of main agent
ac = _get_activity_condition(node_modifier_data)
main_agent.activity = ac
return main_agent
# OTHER NODE TYPES -----
# Get node identifier information
name = node_data.get(pc.NAME)
ns = node_data[pc.NAMESPACE]
ident = node_data.get(pc.IDENTIFIER)
# No ID present, get identifier using the name, namespace
db_refs = None
if not ident:
assert name, "Node must have a name if lacking an identifier."
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 None
db_refs = {'HGNC': hgnc_id}
up_id = _get_up_id(hgnc_id)
if up_id:
db_refs['UP'] = up_id
# FIXME: Look up go ID in ontology lookup service
# FIXME: Look up MESH IDs from name
# FIXME: For now, just use node name
elif ns in ('GOBP', 'MESHPP', 'MESHD'):
db_refs = {}
# 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 == 'EGID':
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)
else:
logger.info('Could not map EGID%s to HGNC.' % name)
name = 'E%s' % name
# CHEBI
elif ns == 'CHEBI':
chebi_id = chebi_name_id.get(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:
print("Unhandled namespace: %s: %s (%s)" % (ns, name, node_data))
# We've already got an identifier, look up other identifiers if necessary
else:
# Get the name, overwriting existing name if necessary
if ns == 'HGNC':
name = hgnc_client.get_hgnc_name(ident)
db_refs = {'HGNC': ident}
up_id = _get_up_id(ident)
if up_id:
db_refs['UP'] = up_id
elif ns == 'UP':
db_refs = {'UP': ident}
name = uniprot_client.get_gene_name(ident)
assert name
if uniprot_client.is_human(ident):
hgnc_id = hgnc_client.get_hgnc_id(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 in ('MGI', 'RGD'):
raise ValueError('Identifiers for MGI and RGD databases are not '
'currently handled: %s' % node_data)
else:
print("Unhandled namespace with identifier: %s: %s (%s)" %
(ns, name, node_data))
if db_refs is None:
logger.info('Unable to get identifier information for node: %s',
node_data)
return None
# Get modification conditions
mods, muts = _get_all_pmods(node_data)
# Get activity condition
ac = _get_activity_condition(node_modifier_data)
to_loc = _get_translocation_target(node_modifier_data)
# Check for unhandled node modifiers, skip if so
if _has_unhandled_modifiers(node_modifier_data):
return None
# Make the agent
ag = Agent(name, db_refs=db_refs, mods=mods, mutations=muts, activity=ac,
location=to_loc)
return ag
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 is_non_human_protein(bio_ontology, node):
"""Return True if the given ontology node is a non-human protein."""
if bio_ontology.get_ns(node) == 'UP' and \
not uniprot_client.is_human(bio_ontology.get_id(node)):
return True
return False
def standardize_db_refs(db_refs):
"""Return a standardized db refs dict for a given db refs dict.
Parameters
----------
db_refs : dict
A dict of db refs that may not be standardized, i.e., may be
missing an available UP ID corresponding to an existing HGNC ID.
Returns
-------
dict
The db_refs dict with standardized entries.
"""
up_id = db_refs.get('UP')
hgnc_id = db_refs.get('HGNC')
# If we have a UP ID and no HGNC ID, we try to get a gene name,
# and if possible, a HGNC ID from that
if up_id and not hgnc_id and uniprot_client.is_human(up_id):
gene_name = uniprot_client.get_gene_name(up_id, False)
if gene_name:
hgnc_id = hgnc_client.get_hgnc_id(gene_name)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
# Otherwise, if we don't have a UP ID but have an HGNC ID, we try to
# get the UP ID
elif hgnc_id:
# Now get the Uniprot ID for the gene
mapped_up_id = hgnc_client.get_uniprot_id(hgnc_id)
if mapped_up_id:
# If we find an inconsistency, we explain it in an error
# message and fall back on the mapped ID
if up_id and up_id != mapped_up_id:
# We handle a special case here in which mapped_up_id is
# actually a list of UP IDs that we skip and just keep
# the original up_id
if ', ' not in mapped_up_id:
# If we got a proper single protein mapping, we use
# the mapped_up_id to standardize to.
msg = ('Inconsistent groundings UP:%s not equal to '
'UP:%s mapped from HGNC:%s, standardizing to '
'UP:%s' %
(up_id, mapped_up_id, hgnc_id, mapped_up_id))
logger.debug(msg)
db_refs['UP'] = mapped_up_id
# If there is no conflict, we can update the UP entry
else:
db_refs['UP'] = mapped_up_id
# Now try to improve chemical groundings
pc_id = db_refs.get('PUBCHEM')
chebi_id = db_refs.get('CHEBI')
hmdb_id = db_refs.get('HMDB')
mapped_chebi_id = None
mapped_pc_id = None
hmdb_mapped_chebi_id = None
# If we have original PUBCHEM and CHEBI IDs, we always keep those:
if pc_id:
mapped_chebi_id = chebi_client.get_chebi_id_from_pubchem(pc_id)
if mapped_chebi_id and not mapped_chebi_id.startswith('CHEBI:'):
mapped_chebi_id = 'CHEBI:%s' % mapped_chebi_id
if chebi_id:
mapped_pc_id = chebi_client.get_pubchem_id(chebi_id)
if hmdb_id:
hmdb_mapped_chebi_id = chebi_client.get_chebi_id_from_hmdb(hmdb_id)
if hmdb_mapped_chebi_id and \
not hmdb_mapped_chebi_id.startswith('CHEBI:'):
hmdb_mapped_chebi_id = 'CHEBI:%s' % hmdb_mapped_chebi_id
# We always keep originals if both are present but display warnings
# if there are inconsistencies
if pc_id and chebi_id and mapped_pc_id and pc_id != mapped_pc_id:
msg = ('Inconsistent groundings PUBCHEM:%s not equal to '
'PUBCHEM:%s mapped from %s, standardizing to '
'PUBCHEM:%s.' % (pc_id, mapped_pc_id, chebi_id, pc_id))
logger.debug(msg)
elif pc_id and chebi_id and mapped_chebi_id and chebi_id != \
mapped_chebi_id:
msg = ('Inconsistent groundings %s not equal to '
'%s mapped from PUBCHEM:%s, standardizing to '
'%s.' % (chebi_id, mapped_chebi_id, pc_id, chebi_id))
logger.debug(msg)
# If we have PC and not CHEBI but can map to CHEBI, we do that
elif pc_id and not chebi_id and mapped_chebi_id:
db_refs['CHEBI'] = mapped_chebi_id
elif hmdb_id and chebi_id and hmdb_mapped_chebi_id and \
hmdb_mapped_chebi_id != chebi_id:
msg = ('Inconsistent groundings %s not equal to '
'%s mapped from %s, standardizing to '
'%s.' % (chebi_id, hmdb_mapped_chebi_id, hmdb_id, chebi_id))
logger.debug(msg)
elif hmdb_id and not chebi_id and hmdb_mapped_chebi_id:
db_refs['CHEBI'] = hmdb_mapped_chebi_id
# If we have CHEBI and not PC but can map to PC, we do that
elif chebi_id and not pc_id and mapped_pc_id:
db_refs['PUBCHEM'] = mapped_pc_id
# Otherwise there is no useful mapping that we can add and no
# further conflict to resolve.
return db_refs