def add_hms_lincs_xrefs(self):
from indra.databases.lincs_client import LincsClient
lc = LincsClient()
edges = []
for hmsl_id, data in lc._sm_data.items():
if '-' in hmsl_id:
hmsl_base_id, suffix = hmsl_id.split('-')
else:
hmsl_base_id, suffix = hmsl_id, None
if suffix == '999':
continue
refs = lc.get_small_molecule_refs(hmsl_id)
for ref_ns, ref_id in refs.items():
edges.append(
(self.label('HMS-LINCS',
hmsl_base_id), self.label(ref_ns, ref_id), {
'type': 'xref',
'source': 'hms-lincs'
}))
edges.append((self.label(ref_ns, ref_id),
self.label('HMS-LINCS', hmsl_base_id), {
'type': 'xref',
'source': 'hms-lincs'
}))
self.add_edges_from(edges)
def __init__(self, lincs_data):
self._data = lincs_data
self._lc = LincsClient()
# Process all the lines (skipping the header)
self.statements = []
for line in self._data:
self._process_line(line)
return
class TasProcessor(object):
"""A processor for the Target Affinity Spectrum data table."""
def __init__(self, data, affinity_class_limit):
self._data = data
self._lc = LincsClient()
self.affinity_class_limit = affinity_class_limit
self.statements = []
for row in data:
# Skip rows that are above the affinity class limit
if int(row['class_min']) > affinity_class_limit:
continue
self._process_row(row)
return
def _process_row(self, row):
drug = self._extract_drug(row['hms_id'])
prot = self._extract_protein(row['approved_symbol'], row['gene_id'])
ev = self._make_evidence(row['class_min'])
# NOTE: there are several entries in this data set that refer to
# non-human Entrez genes, e.g.
# https://www.ncbi.nlm.nih.gov/gene/3283880
# We skip these for now because resources for Entrez-based
# mappings for non-human genes are not integrated, and would cause
# pre-assembly issues.
if 'HGNC' not in prot.db_refs:
return
self.statements.append(Inhibition(drug, prot, evidence=ev))
def _extract_drug(self, hms_id):
refs = self._lc.get_small_molecule_refs(hms_id)
name = self._lc.get_small_molecule_name(hms_id)
if 'PUBCHEM' in refs:
chebi_id = chebi_client.get_chebi_id_from_pubchem(refs['PUBCHEM'])
if chebi_id:
refs['CHEBI'] = 'CHEBI:%s' % chebi_id
return Agent(name, db_refs=refs)
def _extract_protein(self, name, gene_id):
refs = {'EGID': gene_id}
hgnc_id = hgnc_client.get_hgnc_from_entrez(gene_id)
if hgnc_id is not None:
refs['HGNC'] = hgnc_id
up_id = hgnc_client.get_uniprot_id(hgnc_id)
if up_id:
refs['UP'] = up_id
# If there is a HGNC ID, we standardize the gene name
name = hgnc_client.get_hgnc_name(hgnc_id)
return Agent(name, db_refs=refs)
def _make_evidence(self, class_min):
ev = Evidence(source_api='tas', epistemics={'direct': True},
annotations={'class_min': CLASS_MAP[class_min]})
return ev
def __init__(self, lincs_data):
self._data = lincs_data
self._lc = LincsClient()
# Process all the lines (skipping the header)
self.statements = []
for line in self._data:
self._process_line(line)
return
def __init__(self, data, affinity_class_limit):
self._data = data
self._lc = LincsClient()
self.affinity_class_limit = affinity_class_limit
self.statements = []
for row in data:
# Skip rows that are above the affinity class limit
if int(row['class_min']) > affinity_class_limit:
continue
self._process_row(row)
return
def add_hms_lincs_nodes(self):
from indra.databases.lincs_client import LincsClient
lc = LincsClient()
nodes = []
for hmsl_id, data in lc._sm_data.items():
if '-' in hmsl_id:
hmsl_base_id, suffix = hmsl_id.split('-')
else:
hmsl_base_id, suffix = hmsl_id, None
if suffix == '999':
continue
nodes.append((self.label('HMS-LINCS', hmsl_base_id),
{'name': data['Name']}))
self.add_nodes_from(nodes)
class LincsProcessor(object):
"""Processor for the HMS LINCS drug target dataset.
Parameters
----------
lincs_data : list[dict]
A list of dicts with keys set by the header of the csv, and values from
the data in the csv.
Attributes
----------
statements : list[indra.statements.Statement]
A list of indra statements extracted from the CSV file.
"""
def __init__(self, lincs_data):
self._data = lincs_data
self._lc = LincsClient()
# Process all the lines (skipping the header)
self.statements = []
for line in self._data:
self._process_line(line)
return
def _process_line(self, line):
drug = self._extract_drug(line)
prot = self._extract_protein(line)
if prot is None:
return
evidence = self._make_evidence(line)
self.statements.append(Inhibition(drug, prot, evidence=evidence))
def _extract_drug(self, line):
drug_name = line['Small Molecule Name']
lincs_id = line['Small Molecule HMS LINCS ID']
refs = self._lc.get_small_molecule_refs(lincs_id)
if 'PUBCHEM' in refs:
chebi_id = chebi_client.get_chebi_id_from_pubchem(refs['PUBCHEM'])
if chebi_id:
refs['CHEBI'] = chebi_id
return Agent(drug_name, db_refs=refs)
def _extract_protein(self, line):
# Extract key information from the lines.
prot_name = line['Protein Name']
prot_id = line['Protein HMS LINCS ID']
# Get available db-refs.
db_refs = {}
if prot_id:
db_refs.update(self._lc.get_protein_refs(prot_id))
# Since the resource only gives us an UP ID (not HGNC), we
# try to get that and standardize the name to the gene name
up_id = db_refs.get('UP')
if up_id:
hgnc_id = uniprot_client.get_hgnc_id(up_id)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
prot_name = hgnc_client.get_hgnc_name(hgnc_id)
else:
gene_name = uniprot_client.get_gene_name(up_id)
if gene_name:
prot_name = gene_name
# In some cases lines are missing protein information in which
# case we return None
else:
return None
# Create the agent.
return Agent(prot_name, db_refs=db_refs)
def _make_evidence(self, line):
ev_list = []
key_refs = line['Key References'].split(';')
generic_notes = {
'is_nominal': line['Is Nominal'],
'effective_concentration': line['Effective Concentration']
}
patt = re.compile('(?:pmid|pubmed\s+id):\s+(\d+)', re.IGNORECASE)
for ref in key_refs:
# Only extracting pmids, but there is generally more info available.
m = patt.search(ref)
if m is None:
pmid = None
else:
pmid = m.groups()[0]
annotations = {'reference': ref}
annotations.update(generic_notes)
ev = Evidence('lincs_drug', pmid=pmid, annotations=annotations,
epistemics={'direct': True})
ev_list.append(ev)
return ev_list
from __future__ import absolute_import, print_function, unicode_literals
import unittest
from nose.plugins.attrib import attr
from indra.databases.lincs_client import get_drug_target_data, LincsClient
lc = LincsClient()
@attr('webservice')
@unittest.skip('LINCS web service very unreliable.')
def test_get_drug_target_data():
data_list = get_drug_target_data()
assert len(data_list) > 100, len(data_list)
def test_get_protein_refs():
prot_refs = lc.get_protein_refs('200020')
assert prot_refs.get('UP') == 'P00519'
assert prot_refs.get('EGID') == '25'
assert prot_refs.get('HMS-LINCS') == '200020'
def test_get_sm_name():
sm_name = lc.get_small_molecule_name('10001')
assert sm_name == 'Seliciclib', sm_name
def test_get_sm_refs():
sm_refs = lc.get_small_molecule_refs('10001')
assert sm_refs.get('HMS-LINCS') == '10001', sm_refs
class LincsProcessor(object):
"""Processor for the HMS LINCS drug target dataset.
Parameters
----------
lincs_data : list[dict]
A list of dicts with keys set by the header of the csv, and values from
the data in the csv.
Attributes
----------
statements : list[indra.statements.Statement]
A list of indra statements extracted from the CSV file.
"""
def __init__(self, lincs_data):
self._data = lincs_data
self._lc = LincsClient()
# Process all the lines (skipping the header)
self.statements = []
for line in self._data:
self._process_line(line)
return
def _process_line(self, line):
drug = self._extract_drug(line)
prot = self._extract_protein(line)
if prot is None:
return
evidence = self._make_evidence(line)
self.statements.append(Inhibition(drug, prot, evidence=evidence))
def _extract_drug(self, line):
drug_name = line['Small Molecule Name']
lincs_id = line['Small Molecule HMS LINCS ID']
refs = self._lc.get_small_molecule_refs(lincs_id)
if 'PUBCHEM' in refs:
chebi_id = chebi_client.get_chebi_id_from_pubchem(refs['PUBCHEM'])
if chebi_id:
refs['CHEBI'] = 'CHEBI:%s' % chebi_id
return Agent(drug_name, db_refs=refs)
def _extract_protein(self, line):
# Extract key information from the lines.
prot_name = line['Protein Name']
prot_id = line['Protein HMS LINCS ID']
# Get available db-refs.
db_refs = {}
if prot_id:
db_refs.update(self._lc.get_protein_refs(prot_id))
# Since the resource only gives us an UP ID (not HGNC), we
# try to get that and standardize the name to the gene name
up_id = db_refs.get('UP')
if up_id:
gene_name = uniprot_client.get_gene_name(up_id)
if gene_name:
prot_name = gene_name
hgnc_id = hgnc_client.get_hgnc_id(gene_name)
if hgnc_id:
db_refs['HGNC'] = hgnc_id
# In some cases lines are missing protein information in which
# case we return None
else:
return None
# Create the agent.
return Agent(prot_name, db_refs=db_refs)
def _make_evidence(self, line):
ev_list = []
key_refs = line['Key References'].split(';')
generic_notes = {
'is_nominal': line['Is Nominal'],
'effective_concentration': line['Effective Concentration']
}
patt = re.compile('(?:pmid|pubmed\s+id):\s+(\d+)', re.IGNORECASE)
for ref in key_refs:
# Only extracting pmids, but there is generally more info available.
m = patt.search(ref)
if m is None:
pmid = None
else:
pmid = m.groups()[0]
annotations = {'reference': ref}
annotations.update(generic_notes)
ev = Evidence('lincs_drug', pmid=pmid, annotations=annotations,
epistemics={'direct': True})
ev_list.append(ev)
return ev_list