def make_arg_parser():
arg_parser = argparse.ArgumentParser(description='validate_kg2_util_curies_urls_categories.py: ' +
'checks the file `kg2_util.py` for correctness for its CURIE IDs, Base URLs, and biolink categories.')
arg_parser.add_argument('curiesToURLsMapFile', type=str)
arg_parser.add_argument('biolinkModelURL', type=str)
arg_parser.add_argument('biolinkModelLocalFile', type=str)
return arg_parser
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')}
arg_parser.add_argument('curiesToURLsMapFile', type=str)
arg_parser.add_argument('umls2rdfConfFile', type=str)
arg_parser.add_argument('biolinkModelURL', type=str)
arg_parser.add_argument('biolinkModelLocalFile', type=str)
return arg_parser
args = make_arg_parser().parse_args()
owl_load_inventory_file_name = args.owlLoadInventoryFile
curies_to_urls_map_file_name = args.curiesToURLsMapFile
umls2rdf_conf_file_name = args.umls2rdfConfFile
biolink_model_url = args.biolinkModelURL
biolink_model_file_name = args.biolinkModelLocalFile
owl_load_inventory_data = kg2_util.safe_load_yaml_from_string(
kg2_util.read_file_to_string(owl_load_inventory_file_name))
umls_ttl_files = None
with open(umls2rdf_conf_file_name, 'r') as umls2rdf_conf_file:
umls_ttl_files = {
line.split(",")[1]
for line in umls2rdf_conf_file.read().splitlines()
if len(line) > 0 and not line.startswith('#')
}
umls2rdf_conf_file.close()
umls_ttl_files.add('umls-semantictypes.ttl')
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(
if __name__ == '__main__':
args = make_arg_parser().parse_args()
predicate_remap_file_name = args.predicateRemapYaml
curies_to_uri_lal_file_name = args.curiesToURILALFile
input_file_name = args.inputFileJson
output_file_name = args.outputFileJson
test_mode = args.test
drop_negated = args.drop_negated
drop_self_edges_except = args.drop_self_edges_except
if drop_self_edges_except is not None:
assert type(drop_self_edges_except) == str
drop_self_edges_except = set(drop_self_edges_except.split(','))
predicate_remap_config = kg2_util.safe_load_yaml_from_string(
kg2_util.read_file_to_string(predicate_remap_file_name))
curies_to_uri_map = kg2_util.make_curies_to_uri_map(
curies_to_uri_lal_file_name)
graph = kg2_util.load_json(input_file_name)
edge_keys = set()
new_edges = dict()
relation_curies_not_in_config = set()
record_of_relation_curie_occurrences = {
relation_curie: False
for relation_curie in predicate_remap_config.keys()
}
command_set = {'delete', 'keep', 'invert', 'rename'}
for relation_curie, command in predicate_remap_config.items():
assert len(command) == 1
assert next(iter(command.keys())) in command_set
relation_curies_not_in_nodes = set()
'validate_curies_to_categories.py: checks the file `curies-to-categories.yaml` for correctness.'
)
arg_parser.add_argument('curiesToCategoriesFile', type=str)
arg_parser.add_argument('curiesToURLsMapFile', type=str)
arg_parser.add_argument('biolinkModelURL', type=str)
arg_parser.add_argument('biolinkModelLocalFile', type=str)
return arg_parser
args = make_arg_parser().parse_args()
curies_to_categories_file_name = args.curiesToCategoriesFile
curies_to_urls_map_file_name = args.curiesToURLsMapFile
biolink_model_url = args.biolinkModelURL
biolink_model_file_name = args.biolinkModelLocalFile
curies_to_categories_data = kg2_util.safe_load_yaml_from_string(
kg2_util.read_file_to_string(curies_to_categories_file_name))
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 = {
next(iter(listitem.keys()))
for listitem in curies_to_url_map_data['use_for_bidirectional_mapping']
}
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)
for prefix in curies_to_categories_data['prefix-mappings'].keys():
assert prefix in curies_to_url_map_data_bidir, prefix
'checks the file `kg2_util.py` for correctness for its CURIE IDs, Base URLs, and biolink categories.'
)
arg_parser.add_argument('curiesToURLsMapFile', type=str)
arg_parser.add_argument('biolinkModelURL', type=str)
arg_parser.add_argument('biolinkModelLocalFile', type=str)
return arg_parser
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()
}
try:
biolink_to_external_mappings['biolink:subclass_of']['narrow_mappings'].remove("umls:rb")
biolink_to_external_mappings['biolink:superclass_of']['narrow_mappings'].remove("umls:rb")
except ValueError:
print('UMLS:RB work around no longer necessary')
return biolink_to_external_mappings, biolink_mixins
args = make_arg_parser().parse_args()
curies_to_urls_map_file_name = args.curiesToURLsMapFile
predicate_remap_file_name = args.predicateRemapFile
biolink_model_url = args.biolinkModelYamlURL
biolink_model_file_name = args.biolinkModelYamlLocalFile
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 = {next(iter(listitem.keys(
))) for listitem in curies_to_url_map_data['use_for_bidirectional_mapping']}
kg2_util.download_file_if_not_exist_locally(
biolink_model_url, biolink_model_file_name)
biolink_model = kg2_util.safe_load_yaml_from_string(
kg2_util.read_file_to_string(biolink_model_file_name))
mapping_hierarchy = ["exact_mappings", "close_mappings", "narrow_mappings", "broad_mappings", "related_mappings"] # TODO: determine correct order of mappings
biolink_to_external_mappings, biolink_mixins = create_biolink_to_external_mappings(
biolink_model, mapping_hierarchy)
external_to_biolink_mappings = dict()
arg_parser.add_argument('outputFile', type=str)
return arg_parser
# --------------- main starts here -------------------
if __name__ == '__main__':
delete_ontobio_cachier_caches()
args = make_arg_parser().parse_args()
curies_to_categories_file_name = args.categoriesFile
curies_to_uri_lal_file_name = args.curiesToURILALFile
owl_load_inventory_file = args.owlLoadInventoryFile
output_file = args.outputFile
test_mode = args.test
curies_to_categories = kg2_util.safe_load_yaml_from_string(
kg2_util.read_file_to_string(curies_to_categories_file_name))
curies_to_uri_map = kg2_util.make_curies_to_uri_map(
curies_to_uri_lal_file_name)
uri_to_curie_shortener = make_uri_to_curie_shortener(curies_to_uri_map)
map_category_label_to_iri = functools.partial(
kg2_util.convert_biolink_category_to_iri,
biolink_category_base_iri=kg2_util.BIOLINK_CATEGORY_BASE_IRI)
owl_urls_and_files = tuple(
kg2_util.safe_load_yaml_from_string(
kg2_util.read_file_to_string(owl_load_inventory_file)))
make_kg2(curies_to_categories, uri_to_curie_shortener,
map_category_label_to_iri, owl_urls_and_files, output_file,
test_mode)
def make_arg_parser():
arg_parser = argparse.ArgumentParser(description='validate_curies_to_categories.py: checks the file `curies-to-categories.yaml` for correctness.')
arg_parser.add_argument('curiesToCategoriesFile', type=str)
arg_parser.add_argument('curiesToURLsMapFile', type=str)
arg_parser.add_argument('biolinkModelOWLURL', type=str)
arg_parser.add_argument('biolinkModelOWLLocalFile', type=str)
return arg_parser
args = make_arg_parser().parse_args()
curies_to_categories_file_name = args.curiesToCategoriesFile
curies_to_urls_map_file_name = args.curiesToURLsMapFile
biolink_model_url = args.biolinkModelOWLURL
biolink_model_file_name = args.biolinkModelOWLLocalFile
curies_to_categories_data = kg2_util.safe_load_yaml_from_string(kg2_util.read_file_to_string(curies_to_categories_file_name))
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 = {next(iter(listitem.keys())) for listitem in curies_to_url_map_data['use_for_bidirectional_mapping']}
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)
for prefix in curies_to_categories_data['prefix-mappings'].keys():
assert prefix in curies_to_url_map_data_bidir, prefix
for curie_id in curies_to_categories_data['term-mappings'].keys():
prefix = curie_id.split(':')[0]
assert prefix in curies_to_url_map_data_bidir, prefix
categories_to_check = list(curies_to_categories_data['prefix-mappings'].values()) +\