def make_kg2(curies_to_categories: dict,
uri_to_curie_shortener: callable,
map_category_label_to_iri: callable,
owl_urls_and_files: tuple,
output_file_name: str,
test_mode: bool = False):
owl_file_information_dict_list = []
# for each OWL file (or URL for an OWL file) described in the YAML config file...
for ont_source_info_dict in owl_urls_and_files:
if ont_source_info_dict['download']:
# get the OWL file onto the local file system and get a full path to it
local_file_name = kg2_util.download_file_if_not_exist_locally(
ont_source_info_dict['url'], ont_source_info_dict['file'])
else:
local_file_name = ont_source_info_dict['file']
assert os.path.exists(ont_source_info_dict['file'])
# load the OWL file dadta into an ontobio.ontol.Ontology data structure and information dictionary
[ont, metadata_dict] = load_owl_file_return_ontology_and_metadata(
local_file_name, ont_source_info_dict['url'],
ont_source_info_dict['title'])
metadata_dict['ontology'] = ont
owl_file_information_dict_list.append(metadata_dict)
kg2_util.log_message('Calling make_nodes_dict_from_ontologies_list')
nodes_dict = make_nodes_dict_from_ontologies_list(
owl_file_information_dict_list, curies_to_categories,
uri_to_curie_shortener, map_category_label_to_iri)
kg2_util.log_message('Calling make_map_of_node_ontology_ids_to_curie_ids')
map_of_node_ontology_ids_to_curie_ids = make_map_of_node_ontology_ids_to_curie_ids(
nodes_dict)
kg2_util.log_message('Calling get_rels_dict')
# get a dictionary of all relationships including xrefs as relationships
all_rels_dict = get_rels_dict(nodes_dict, owl_file_information_dict_list,
uri_to_curie_shortener,
map_of_node_ontology_ids_to_curie_ids)
kg2_dict = dict()
kg2_dict['edges'] = [rel_dict for rel_dict in all_rels_dict.values()]
kg2_util.log_message('Number of edges: ' + str(len(kg2_dict['edges'])))
kg2_dict['nodes'] = list(nodes_dict.values())
kg2_util.log_message('Number of nodes: ' + str(len(kg2_dict['nodes'])))
del nodes_dict
# delete xrefs from all_nodes_dict
for node_dict in kg2_dict['nodes']:
del node_dict['xrefs']
del node_dict['ontology node ids']
kg2_util.log_message('Saving JSON file')
kg2_util.save_json(kg2_dict, output_file_name, test_mode)
args = make_arg_parser().parse_args()
biolink_model_url = args.biolinkModelURL
biolink_model_file_name = args.biolinkModelLocalFile
curies_to_urls_map_file_name = args.curiesToURLsMapFile
iri_shortener = kg2_util.make_uri_to_curie_shortener(kg2_util.make_curies_to_uri_map(kg2_util.read_file_to_string(curies_to_urls_map_file_name),
kg2_util.IDMapperType.CONTRACT))
curies_to_url_map_data = kg2_util.safe_load_yaml_from_string(kg2_util.read_file_to_string(curies_to_urls_map_file_name))
curies_to_url_map_data_bidir = {key: listitem[key] for listitem in curies_to_url_map_data['use_for_bidirectional_mapping'] for key in listitem.keys()}
curies_to_url_map_data_cont = {key: listitem[key] for listitem in curies_to_url_map_data['use_for_contraction_only'] for key in listitem.keys()}
kg2_util.download_file_if_not_exist_locally(biolink_model_url, biolink_model_file_name)
biolink_ont = kg2_util.make_ontology_from_local_file(biolink_model_file_name)
biolink_categories_ontology_depths = kg2_util.get_biolink_categories_ontology_depths(biolink_ont)
biolink_edge_labels = {url.replace(kg2_util.BASE_URL_BIOLINK_META, '') for url in
biolink_ont.children(kg2_util.BASE_URL_BIOLINK_META + 'SlotDefinition')}
for variable_name in dir(kg2_util):
variable_value = getattr(kg2_util, variable_name)
if variable_name.startswith('CURIE_PREFIX_'):
assert variable_value in curies_to_url_map_data_bidir, variable_name
elif variable_name.startswith('BASE_URL_'):
url_str = variable_value
curie = iri_shortener(url_str)
assert curie is not None, url_str
elif variable_name.startswith('BIOLINK_CATEGORY_'):