def test_prefixes():
assert contract_uri(bp_iri) == [bp_id]
assert expand_uri(bp_id) == bp_iri
assert contract_uri("FAKE", strict=False) == []
try:
contract_uri("FAKE", strict=True)
except NoPrefix as e:
pass
else:
assert False
def contract(uri) -> str:
"""
We sort the curies to ensure that we take the same item every time
"""
curies = contract_uri(str(uri), cmaps=cmaps)
if len(curies) > 0:
curies.sort()
return curies[0]
return None
def test_prefixes_cmaps():
cmaps = [ {'GO': 'http://purl.obolibrary.org/obo/GO_'},
{'OBO': 'http://purl.obolibrary.org/obo/'}
]
assert contract_uri(bp_iri, cmaps) == [bp_id]
all_curies = contract_uri(bp_iri, cmaps, shortest=False)
assert len(all_curies) == 2
assert obo_bp_id in all_curies
assert bp_id in all_curies
assert expand_uri(bp_id, cmaps) == bp_iri
assert expand_uri(obo_bp_id, cmaps) == bp_iri
assert contract_uri("FAKE", cmaps, strict=False) == []
try:
contract_uri("FAKE", cmaps, strict=True)
except NoPrefix as e:
pass
else:
assert False
def get_descendants(graph: Graph,
node: str,
edge: Optional[URIRef] = RDFS['subClassOf'],
reflexive: Optional[bool] = True) -> Set[str]:
nodes = set()
node = URIRef(expand_uri(node, strict=True))
for sub in graph.transitive_subjects(edge, node):
if not reflexive and node == sub:
continue
if isinstance(sub, Literal):
continue
nodes.add(contract_uri(str(sub), strict=True)[0])
return nodes
def contract_uri(self, iri) -> str:
"""Contract a given IRI.
Contract a given IRI, with special parsing and transformations
depending on the nature of the IRI.
Args:
iri: IRI as string
Returns:
str.
"""
curie = ""
if 'http://www.genenames.org/cgi-bin/gene_symbol_report?match=' in iri:
identifier = iri.split('=')[-1]
if identifier in self.gene_info_map:
curie = f"NCBIGene:{self.gene_info_map[identifier]['NCBI']}"
else:
[curie] = contract_uri(iri, cmaps=[CUSTOM_CMAP])
else:
if self.is_iri(iri):
curie = contract_uri(iri)
if curie:
curie = curie[0]
else:
curie = contract_uri(iri, cmaps=[CUSTOM_CMAP])
if curie:
curie = curie[0]
else:
curie = iri
elif self.is_curie(iri):
curie = iri
else:
curie = f":{iri}"
return curie
def get_ancestors(graph: Graph,
node: str,
edge: Optional[URIRef] = RDFS['subClassOf'],
root: Optional[str] = None,
reflexive: Optional[bool] = True) -> Set[str]:
nodes = set()
root_seen = {}
node = URIRef(expand_uri(node, strict=True))
if root is not None:
root = URIRef(expand_uri(root, strict=True))
root_seen = {root: 1}
for obj in graph.transitive_objects(node, edge, root_seen):
if isinstance(obj, Literal) or isinstance(obj, BNode):
continue
if not reflexive and node == obj:
continue
nodes.add(contract_uri(str(obj), strict=True)[0])
# Add root to graph
if root is not None:
nodes.add(contract_uri(str(root), strict=True)[0])
return nodes
def get_leaf_nodes(graph: Graph,
node: str,
edge: Optional[URIRef] = RDFS['subClassOf']) -> Set[str]:
if not isinstance(node, URIRef):
obj = URIRef(expand_uri(node, strict=True))
else:
obj = node
subjects = list(graph.subjects(edge, obj))
if len(subjects) == 0:
yield contract_uri(str(obj), strict=True)[0]
else:
for subject in subjects:
for leaf in get_leaf_nodes(graph, subject, edge):
yield leaf
def shorten_iri_to_curie(iri: str, curie_to_iri_map: list = []):
if iri.startswith('owl:') or iri.startswith('OIO:'):
return iri
if "/GO/GO%3A" in iri: # hack for fixing issue #410
iri = iri.replace("/GO/GO%3A", "/GO/")
if "/HPO/HP%3A" in iri: # hack for fixing issue #665
iri = iri.replace("/HPO/HP%3A", "/HP/")
curie_list = prefixcommons.contract_uri(iri, curie_to_iri_map)
assert len(curie_list) in [0, 1]
if len(curie_list) == 1:
curie_id = curie_list[0]
else:
curie_id = None
if curie_id is not None:
# deal with IRIs like 'https://identifiers.org/umls/ATC/L01AX02' which get converted to CURIE 'UMLS:ATC/L01AX02'
umls_match = REGEX_UMLS_CURIE.match(curie_id)
if umls_match is not None:
curie_id = umls_match[1] + ':' + umls_match[2]
return curie_id
def contract(uri) -> str:
"""
Contract a URI a CURIE.
We sort the curies to ensure that we take the same item every time.
Parameters
----------
uri: Union[rdflib.term.URIRef, str]
A URI
Returns
-------
str
The CURIE
"""
curies = contract_uri(str(uri), cmaps=cmaps)
if len(curies) > 0:
curies.sort()
return curies[0]
return None
def shorten_iri_to_curie(iri: str, curie_to_iri_map: list) -> str:
if iri is None:
raise ValueError('cannot shorten an IRI with value None')
curie_list = prefixcommons.contract_uri(iri, curie_to_iri_map)
if len(curie_list) == 0:
return None
if len(curie_list) == 1:
curie_id = curie_list[0]
else:
assert False, "somehow got a list after calling prefixcommons.contract on URI: " + iri + "; list is: " + str(
curie_list)
curie_id = None
# if curie_id is not None:
# # deal with IRIs like 'https://identifiers.org/umls/ATC/L01AX02' which get converted to CURIE 'UMLS:ATC/L01AX02'
# umls_match = REGEX_UMLS_CURIE.match(curie_id)
# if umls_match is not None:
# curie_id = umls_match[1] + ':' + umls_match[2]
return curie_id
def _process_hpo_data(file_path: str) -> Dict[str, List[str]]:
logger.info("loading mondo into memory")
mondo = Graph()
mondo.parse(gzip.open("../data/mondo.owl.gz", 'rb'), format='xml')
logger.info("finished loading mondo")
mondo_merged_lines: List[str] = []
disease_info: Dict[str, List[str]] = {}
if file_path.startswith("http"):
context_manager = closing(requests.get(file_path))
else:
context_manager = open(file_path, "r")
# https://stackoverflow.com/a/35371451
with context_manager as file:
if file_path.startswith("http"):
file = file.content.decode('utf-8').splitlines()
reader = csv.reader(file, delimiter='\t', quotechar='\"')
counter = 0
for row in reader:
try:
(db, num, name, severity, pheno_id, publist, eco, onset,
freq) = row[0:9]
except ValueError:
logger.warning("Too few values in row {}".format(row))
continue
# Align Id prefixes
if db == 'MIM': db = 'OMIM'
if db == 'ORPHA': db = 'Orphanet'
if db == 'ORPHANET': db = 'Orphanet'
disease_id = "{}:{}".format(db, num)
disease_iri = URIRef(expand_uri(disease_id, strict=True))
mondo_curie = None
mondo_iri = None
for subj in mondo.subjects(OWL['equivalentClass'], disease_iri):
curie = contract_uri(str(subj), strict=True)[0]
if curie.startswith('MONDO'):
mondo_curie = curie
mondo_iri = subj
break
if mondo_curie is None:
logger.warn("No mondo id for {}".format(disease_id))
continue
has_omim = False
for obj in mondo.objects(mondo_iri, OWL['equivalentClass']):
try:
curie = contract_uri(str(obj), strict=True)[0]
except NoPrefix:
continue
if curie.startswith('OMIM'):
has_omim = True
# use scigraph instead of the above
# mondo_node = monarch.get_clique_leader(disease_id)
# mondo_curie = mondo_node['id']
if mondo_curie is not None and 'hgnc' in mondo_curie:
# to keep these, likely decipher IDs
# mondo_curie = disease_id
continue
if disease_id.startswith('Orphanet') \
and has_omim is False \
and len(list(mondo.objects(mondo_iri, RDFS['subClassOf']))) > 0:
# disease is a disease group, skip
logger.info(
"{} is a disease group, skipping".format(disease_id))
continue
mondo_merged_lines.append(
(mondo_curie, pheno_id, onset, freq, severity))
counter += 1
if counter % 10000 == 0:
logger.info("processed {} rows".format(counter))
logger.info("processed {} rows".format(counter))
for line in mondo_merged_lines:
key = "{}-{}".format(line[0], line[1])
values = [line[2], line[3], line[4]]
if key in disease_info and disease_info[key] != values:
logger.warning("Metadata for {} and {} mismatch: {} vs {}".format(
line[0], line[1], values, disease_info[key]))
# attempt to merge by collapsing freq, onset, severity
# that is empty in one disease but not another
# conflicts will defer to the disease first inserted
merged_disease_info = disease_info[key]
for index, val in enumerate(values):
if val == disease_info[key][index] \
or val == '' and disease_info[key][index] != '':
continue
elif val != '' and disease_info[key][index] == '':
merged_disease_info[index] = val
else:
logger.warning("Cannot merge {} and {} for {}".format(
values, disease_info[key], line[0]))
else:
disease_info[key] = values
return disease_info
