def main(path, output, model):
if model is not None:
bmt.load(model)
t = JsonTransformer()
t.parse(path)
def curie_to_label(curie:str):
"""
Uses the biolink model toolkit to look up an
element (on the tree rooted at `named thing`
and `related to`) for a given curie. If none
can be found then returns the original curie.
"""
if isinstance(curie, (list, tuple, set)):
return [curie_to_label(c) for c in curie]
elif isinstance(curie, str):
e = bmt.get_by_mapping(curie)
return e if e is not None else curie
else:
return None
for n, attr in t.graph.nodes(data=True):
attr['category'] = curie_to_label(attr.get('category'))
for s, o, attr in t.graph.edges(data=True):
attr['predicate'] = curie_to_label(attr.get('predicate'))
t.save(output)
def test_json_save():
t = JsonTransformer()
t.parse(os.path.join(resource_dir, 'valid.json'))
assert t.graph.number_of_nodes() == 6
assert t.graph.number_of_edges() == 5
t.save(os.path.join(target_dir, 'graph.json'))
assert os.path.exists(os.path.join(target_dir, 'graph.json'))
def main(path, output, model):
if model is not None:
bmt.load(model)
t = JsonTransformer()
t.parse(path)
t = PandasTransformer(t.graph)
t.save(output)
def test_neo_to_graph_upload():
""" loads a neo4j graph from a json file
"""
jt = JsonTransformer()
jt.parse('resources/robodb2.json')
nt = NeoTransformer(jt.graph, host='localhost', port='7474', username='******', password='******')
nt.save_with_unwind()
nt.neo4j_report()
def test_validate_json():
"""
Validate against a valid representative Biolink Model compliant JSON
"""
json_file = os.path.join(resource_dir, 'valid.json')
jt = JsonTransformer()
jt.parse(json_file)
validator = Validator()
e = validator.validate(jt.graph)
assert len(e) == 0
def test_load():
"""
Test for loading into JsonTransformer
"""
json_file = os.path.join(resource_dir, 'semmed/gene.json')
jt = JsonTransformer()
jt.parse(json_file)
edge_list = list(jt.graph.edges(data=True))
assert edge_list[0][-1]['subject'] == 'UMLS:C0948075'
assert edge_list[0][-1]['object'] == 'UMLS:C1290952'
def test_export():
"""
Test export behavior of JsonTransformer
"""
json_file = os.path.join(resource_dir, 'semmed/gene.json')
output_file = os.path.join(target_dir, 'semmeddb_export.json')
jt = JsonTransformer()
jt.parse(json_file)
jt.save(output_file)
assert os.path.isfile(output_file)
def test_neo_to_graph_upload():
""" loads a neo4j graph from a json file
"""
jt = JsonTransformer()
jt.parse('resources/robodb2.json')
nt = NeoTransformer(jt.graph,
uri=DEFAULT_NEO4J_URL,
username=DEFAULT_NEO4J_USERNAME,
password=DEFAULT_NEO4J_PASSWORD)
nt.save()
nt.neo4j_report()
def test_json_load():
t = JsonTransformer()
t.parse(os.path.join(resource_dir, 'valid.json'))
assert t.graph.number_of_nodes() == 6
assert t.graph.number_of_edges() == 5
n = t.graph.nodes['MONDO:0017148']
assert isinstance(n, dict)
assert 'id' in n and n['id'] == 'MONDO:0017148'
assert n['name'] == 'heritable pulmonary arterial hypertension'
assert n['category'][0] == 'biolink:Disease'
data = t.graph.get_edge_data('HGNC:11603', 'MONDO:0017148')
assert len(data.keys()) == 1
data = data.popitem()[1]
assert data['subject'] == 'HGNC:11603'
assert data['object'] == 'MONDO:0017148'
assert data['edge_label'] == 'biolink:related_to'
assert data['relation'] == 'RO:0004013'
from kgx import JsonTransformer, clique_merge
t = JsonTransformer()
t.parse('results/hp.owl')
t.parse('results/mondo.json')
t.parse('results/hgnc.json')
t.parse('results/clinvar.json')
t.parse('results/omim.json')
t.parse('results/hpoa.json')
t.parse('results/orphanet.json')
#t = PandasTransformer(t.graph)
#t.parse('data/semmeddb_edges.csv')
#t.parse('data/semmeddb_nodes.csv')
t.graph = clique_merge(t.graph)
t.save('results/clique_merged.json')
"""
This script prepares the clique_merged.json file for uploading to Neo4j
- Removes nodes that cannot be categorized into the biolink model
- Renames edge labels that don't matche the biolink model to "related_to"
- Transforms into CSV format
"""
from kgx import JsonTransformer, PandasTransformer
import bmt
t = JsonTransformer()
t.parse('results/clique_merged.json')
t = PandasTransformer(t)
G = t.graph
size = len(G)
nodes = []
for n, data in G.nodes(data=True):
data['category'] = [
c for c in data.get('category', []) if bmt.get_class(c) is not None
]
if data['category'] == []:
if 'name' in data:
data['category'] = ['named thing']
else:
nodes.append(n)
G.remove_nodes_from(nodes)
from kgx import JsonTransformer, clique_merge
import sys
path = sys.argv[1]
t = JsonTransformer()
t.parse(path)
t.graph = clique_merge(t.graph)
t.save('clique_merged.json')
Loads all the turtle files with their required ontologies and transforms them to
json. Then loads all these json files, along with the semmeddb edges.csv and
nodes.csv files, into a single NetworkX graph, and performs `clique_merge` on it.
Finally, saves the resulting NetworkX graph as `clique_merged.csv`
"""
from kgx import ObanRdfTransformer2, JsonTransformer, HgncRdfTransformer, RdfOwlTransformer2
from kgx import clique_merge, make_valid_types
t = RdfOwlTransformer2()
t.parse('data/hp.owl')
t = JsonTransformer(t)
t.save('results/hp.json')
t = RdfOwlTransformer2()
t.parse('data/mondo.owl')
t = JsonTransformer(t)
t.save('results/mondo.json')
t = HgncRdfTransformer()
t.parse('data/hgnc.ttl')
t = JsonTransformer(t)
t.save('results/hgnc.json')
t = ObanRdfTransformer2()
t.add_ontology('data/mondo.owl')
t.add_ontology('data/hp.owl')
t.parse('data/orphanet.ttl')
t = JsonTransformer(t)
t.save('results/orphanet.json')
from kgx import ObanRdfTransformer, JsonTransformer, HgncRdfTransformer
from kgx import clique_merge
t = JsonTransformer()
t.parse('hgnc.json')
t.parse('clinvar.json')
t.parse('omim.json')
t.parse('hpoa.json')
t.parse('orphanet.json')
t.save('merged.json')
t.graph = clique_merge(t.graph)
t.save('clique_merged.json')
from kgx import ObanRdfTransformer, JsonTransformer, HgncRdfTransformer
from kgx import clique_merge
t = JsonTransformer()
#t.parse('hgnc.json')
#t.parse('clinvar.json')
#t.parse('omim.json')
#t.parse('hpoa.json')
#t.parse('orphanet.json')
t.parse('semmeddb.json')
t.parse('merged.json')
t.save('merged.json')
t.graph = clique_merge(t.graph)
t.save('clique_merged.json')