def __init__(self, entities):
graph = KnowledgeGraph()
#num_entities = graph.get_num_nodes()
self.entity_list = entities
self.entities = {item: idx for idx, item in enumerate(entities)}
self.relation_list = ['NO_OP', 'DUMMY_RELATION', 'DONE'] + [
record["predicate"] for record in graph.get_predicates()
]
self.relations = {
item: idx
for idx, item in enumerate(self.relation_list)
}
spec = {"id": "neoenv"}
super().__init__(graph, len(self.entities), len(self.relations), spec)
import pandas as pd
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
from reasoner.knowledge_graph.SemmedDbTools import SemmedDbTools
sdb_tools = SemmedDbTools()
kg = KnowledgeGraph()
sem2type = {
'dsyn': 'Disease',
'neop': 'Disease',
'fndg': 'Disease',
'sosy': 'Symptom',
'tisu': 'Tissue',
'bpoc': 'Tissue',
'blor': 'Tissue',
'cell': 'Cell',
'moft': 'Pathway',
'celf': 'Pathway'
}
terms = sdb_tools.get_terms()
for term in terms:
semtypes = set(term['semtype'].split(','))
st_raw = semtypes.copy()
for st in st_raw:
if st in sem2type:
semtypes.add(sem2type[term['semtype']])
kg.add_umls_term("UMLS:" + term['cui'], term['name'], semtypes)
triples = sdb_tools.get_triples()
for triple in triples:
from reasoner.knowledge_graph.ChemblTools import ChemblTools
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
from reasoner.knowledge_graph.umls.UmlsQuery import UmlsQuery
kg = KnowledgeGraph()
uq = UmlsQuery()
ct = ChemblTools()
chembl_ids = kg.get_drug_chembl_ids()
for chembl_id in chembl_ids:
indications = ct.get_indication(chembl_id.replace('CHEMBL:', ''))
for row in indications:
result = uq.mesh2cui(row['mesh_id'])
if result:
kg.add_indication_relation(chembl_id, 'UMLS:' + result[0]['cui'])
map_file = "../data/knowledge_graph/id_maps/umls2uberon.csv"
id_map_df = pandas.read_csv(map_file)
id_map = {}
for index, row in id_map_df.iterrows():
id_map[row['uberon_id']] = row['umls_id']
## load ontology
owlready2.onto_path.append("/data/owlready")
onto = owlready2.get_ontology("http://purl.obolibrary.org/obo/uberon.owl")
onto.load()
obo = onto.get_namespace("http://purl.obolibrary.org/obo/")
ontology_classes = obo.UBERON_0001062.descendants()
ontology_classes.add(obo.UBERON_0001062)
kg = KnowledgeGraph()
uq = UmlsQuery()
# add terms
for current_class in ontology_classes:
current_id = current_class.name.replace('_', ':')
if current_id in id_map:
cui = id_map[current_id]
umls_result = uq.cui2bestname(cui)
if umls_result:
name = umls_result[0]['name']
else:
name = current_class.label
else:
name = current_class.label
cui = None
import pandas as pd
from reasoner.knowledge_graph.ChemblTools import ChemblTools
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
from reasoner.knowledge_graph.umls.UmlsQuery import UmlsQuery
cop_file = '../data/cop/cop_list_full.csv'
outfile = '../data/knowledge_graph/ready_to_load/diseases.csv'
cop = pd.read_csv(cop_file)
unique_diseases = cop.Condition.unique()
kg = KnowledgeGraph()
uq = UmlsQuery()
ct = ChemblTools()
added = set()
disease_data = pd.DataFrame(columns=["cui", "name", "hpo_id", "mesh_id"])
for term in unique_diseases:
result = uq.meshterm2cui(term)
if result and result[0]['mesh_id'] not in added:
cui = result[0]['cui']
mesh_id = result[0]['mesh_id']
name = uq.cui2bestname(cui)[0]['name']
hpo_result = uq.cui2hpo(cui)
if hpo_result:
hpo_id = hpo_result[0]['hpo_id']
else:
hpo_id = ''
added.add(result[0]['mesh_id'])
disease_data = disease_data.append(
{
import pandas
from reasoner.knowledge_graph.ChemblTools import ChemblTools
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
cop_drugs_file = '../data/neo4j/cop_chembl_ids_curated.csv'
cop_drugs = pandas.read_csv(cop_drugs_file)
ct = ChemblTools()
kg = KnowledgeGraph()
for index, row in cop_drugs.iterrows():
cui = row['cui']
if not pandas.isnull(row['chembl_id']):
chembl_id = row['chembl_id']
else:
chembl_id = None
if not pandas.isnull(row['chebi_id']):
chebi_id = row['chebi_id']
else:
chebi_id = None
if not pandas.isnull(row['drugbank_id']):
drugbank_id = row['drugbank_id']
else:
drugbank_id = None
# print(cui, chembl_id, chebi_id, drugbank_id)
kg.add_drug(cui, chembl_id, chebi_id, drugbank_id)
# add targets
if chembl_id is not None:
targets = ct.get_targets(chembl_id)
for target in targets:
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
from reasoner.knowledge_graph.umls.UmlsQuery import UmlsQuery
uq = UmlsQuery()
kg = KnowledgeGraph()
result = uq.get_snomed_finding_sites()
for row in result:
kg.add_disease_finding_site_relation(row['disease_cui'],
row['location_cui'])
id_map_df = pandas.read_csv(map_file)
id_map = {}
for index, row in id_map_df.iterrows():
id_map[row['symp_id']] = row['umls_id']
## load ontology
owlready2.onto_path.append("/data/owlready")
onto = owlready2.get_ontology("http://purl.obolibrary.org/obo/symp.owl")
onto.load()
obo = onto.get_namespace("http://purl.obolibrary.org/obo/")
###
ontology_classes = obo.SYMP_0000462.descendants()
ontology_classes.add(obo.SYMP_0000462)
kg = KnowledgeGraph()
uq = UmlsQuery()
# add terms
for current_class in ontology_classes:
current_id = current_class.name.replace('_', ':')
if current_id in id_map:
cui = id_map[current_id]
umls_result = uq.cui2bestname(cui)
if umls_result:
name = umls_result[0]['name']
else:
name = current_class.label
else:
name = current_class.label
cui = None
id_map[row['chebi_id']] = 'UMLS:' + row['cui']
## load ontology
owlready2.onto_path.append("/data/owlready")
onto = owlready2.get_ontology(
"file:////home/mwawer/src/reasoner/data/knowledge_graph/primary/chebi.owl")
onto.load()
obo = onto.get_namespace("http://purl.obolibrary.org/obo/")
###
ontology_classes = obo.CHEBI_24431.descendants() # chemical entity
ontology_classes |= obo.CHEBI_50906.descendants() # role
ontology_classes.add(obo.CHEBI_24431)
ontology_classes.add(obo.CHEBI_50906)
kg = KnowledgeGraph()
uq = UmlsQuery()
# add terms
for current_class in ontology_classes:
current_id = current_class.name.replace('_', ':')
if current_id in id_map:
cui = id_map[current_id]
umls_result = uq.cui2bestname(cui)
if umls_result:
name = umls_result[0]['name']
else:
name = current_class.label
else:
name = current_class.label
cui = None
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
outfile = '../data/neo4j/all_chebi_terms.txt'
kg = KnowledgeGraph()
chebi_terms = [record['chebi_id'] for record in kg.get_chebi_terms()]
with open(outfile, 'w') as f:
for term in chebi_terms:
f.write("%s\n" % term)
import pandas as pd
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
pathway_file = '../data/knowledge_graph/ready_to_load/pathways.csv'
protein2pathway_file = '../data/knowledge_graph/ready_to_load/protein_to_pathways.csv'
kg = KnowledgeGraph()
pathways = pd.read_csv(pathway_file)
pathways.fillna('', inplace=True)
for index, row in pathways.iterrows():
kg.add_pathway(row['go_id'], row['name'], row['cui'])
protein2pathway = pd.read_csv(protein2pathway_file)
protein2pathway.fillna('', inplace=True)
for index, row in protein2pathway.iterrows():
if row['db'] == 'UniProtKB':
kg.add_protein_pathway_relation(row['uniprot_id'], row['go_id'],
row['evidence_code'])
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
from reasoner.knowledge_graph.umls.UmlsQuery import UmlsQuery
kg = KnowledgeGraph()
uq = UmlsQuery()
cuis = kg.get_cuis()
for cui in cuis:
semtypes = uq.get_semtype(cui)
for record in semtypes:
kg.set_semtype(cui, record['type_name'].decode())
import csv
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
outfile = 'data/knowledge_graph/export/edgelist.txt'
kg = KnowledgeGraph()
result = kg.get_edgelist()
edgelist = []
for record in result:
edgelist.append([record['start'], record['end']])
with open(outfile, 'w') as f:
writer = csv.writer(f, delimiter='\t')
writer.writerows(edgelist)
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
from reasoner.knowledge_graph.ChemblTools import ChemblTools
kg = KnowledgeGraph()
chembl = ChemblTools()
drugs = kg.get_drug_chembl_ids()
for chembl_id in drugs:
targets = chembl.get_targets(chembl_id)
for target in targets:
if target['component_type'] == 'PROTEIN' and target['db_source'] == 'SWISS-PROT':
kg.add_drug_target_relation(
drug_chembl_id='CHEMBL:' + target['chembl_id'],
target_type='Protein',
target_id_type='uniprot_id',
target_id='UNIPROT:' + target['accession'],
activity_value=float(target['standard_value']),
activity_type=target['standard_type'],
activity_unit=target['standard_units'])
trials = trials + 1
continue
# ontology_classes = obo.HP_0031797.descendants()
# ontology_classes = obo.HP_0012823.descendants()
# ontology_classes = obo.HP_0040279.descendants()
# ontology_classes = obo.HP_0000005.descendants()
# ontology_classes = obo.UPHENO_0001001.descendants()
# ontology_classes.add(obo.HP_0000001)
# ontology_classes.add(obo.HP_0031797)
# ontology_classes.add(obo.HP_0012823)
# ontology_classes.add(obo.HP_0040279)
# ontology_classes.add(obo.HP_0000005)
# ontology_classes.add(obo.UPHENO_0001001)
kg = KnowledgeGraph()
uq = UmlsQuery()
# add terms
for current_class in ontology_classes:
current_id = current_class.name.replace('_', ':')
umls_result = uq.hpo2cui(current_id)
if umls_result:
name = umls_result[0]['name']
cui = 'UMLS:' + umls_result[0]['cui']
else:
name = current_class.label
cui = None
kg.add_hpo_term(current_id, name, cui)
# add relations
import pandas as pd
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
disease_file = '../data/knowledge_graph/ready_to_load/diseases.csv'
disease = pd.read_csv(disease_file)
disease.fillna('', inplace=True)
kg = KnowledgeGraph()
for index, row in disease.iterrows():
if row['hpo_id'] == '':
kg.add_disease(row['cui'], row['name'], row['mesh_id'])
else:
kg.add_disease(row['cui'], row['name'], row['mesh_id'], row['hpo_id'])
np.random.seed(439572)
def get_preferred_label(labels):
if 'Drug' in labels:
return ('Drug')
elif 'Disease' in labels:
return ('Disease')
else:
return ("ChebiTerm")
outfolder = 'translator_test_1'
kg = KnowledgeGraph()
result = kg.query("""
MATCH path = (dr:Drug)-[:HAS_ROLE]->(t:ChebiTerm)--(dis:Disease)--(dr)
UNWIND relationships(path) as r
RETURN startNode(r) as start, r, endNode(r) as end
""")
graph_triples = []
target_triples = []
for record in result:
start_term = get_preferred_label(record['start'].labels) + '_' + re.sub(
r'[ ,\'-]', "", record['start']['name'])
end_term = get_preferred_label(record['end'].labels) + '_' + re.sub(
r'[ ,\'-]', "", record['end']['name'])
relation = record['r'].type
from reasoner.knowledge_graph.KnowledgeGraph import KnowledgeGraph
gene_file = '../data/knowledge_graph/ready_to_load/hgnc_genes_proteins.csv'
kg = KnowledgeGraph()
with open(gene_file) as f:
first_line = f.readline()
for line in f:
items = line.strip('\n').split(',')
kg.add_gene(hgnc_id=items[0], hgnc_symbol=items[1], entrez_id=items[3], name=items[2])
kg.add_protein(uniprot_id=items[4], name=items[2])
kg.add_gene_product_relation(hgnc_id=items[0], uniprot_id=items[4])
owlready2.onto_path.append("/data/owlready")
onto = owlready2.get_ontology("http://purl.obolibrary.org/obo/cl.owl")
onto.load()
obo = onto.get_namespace("http://purl.obolibrary.org/obo/")
###
ontology_classes = obo.CL_0000003.descendants()
ontology_classes |= obo.CL_0001034.descendants()
ontology_classes |= obo.CL_0001061.descendants()
ontology_classes.add(obo.CL_0000000)
# ontology_classes.add(obo.CL_0000003)
# ontology_classes.add(obo.CL_0001034)
# ontology_classes.add(obo.CL_0001061)
kg = KnowledgeGraph()
uq = UmlsQuery()
# add terms
for current_class in ontology_classes:
current_id = current_class.name.replace('_', ':')
if current_id in id_map:
cui = id_map[current_id]
umls_result = uq.cui2bestname(cui)
if umls_result:
name = umls_result[0]['name']
else:
name = current_class.label
else:
name = current_class.label
cui = None