def update(
index,
input_prefix,
output_prefix,
input_value=None,
subject_category=None,
object_category=None,
predicate=None
):
try:
index = int(index)
except:
raise Exception('Index "{}" is not a string'.format(index))
uri = _uri_pattern.format(
input_prefix=input_prefix,
output_prefix=output_prefix,
input_value=input_value
)
response = requests.get(uri)
if response.ok:
data = response.json()
else:
raise Exception('URI {} responsed with status code {}'.format(uri, response.status_code))
mappings = data['data']
if len(mappings) <= index:
raise Exception('Tried to access index {} when there are only {} many'.format(index, len(mappings)))
mapping = mappings[index]
if len(mapping) < 1:
raise Exception('The semantic query at {} is empty'.format(index))
subject_id = safe_get(mapping[0], 'input')
object_id = safe_get(mapping[-1], 'output', 'object', 'id')
object_secondary_id = safe_get(mapping[-1], 'output', 'object', 'secondary-id')
e = {
'subject_id' : safe_get(mapping[0], 'input'),
'object_id' : safe_get(mapping[-1], 'output', 'object', 'id'),
'predicate' : predicate
}
return {'apis' : mapping, 'example' : e, 'uri' : uri}
def get_statement_details(statementId, keywords=None, offset=None, size=None):
if ':' not in statementId:
return BeaconStatementWithDetails()
p = statementId.split(':')
if len(p) != 5:
return BeaconStatementWithDetails()
subject_id = '{}:{}'.format(p[0], p[1])
predicate = '{}'.format(p[2])
object_id = '{}:{}'.format(p[3], p[4])
data = crawler.crawl(subject_id)
for category, assocations in data.items():
for a in assocations:
object_match = object_id == simplify_curie(
safe_get(a, 'object', 'id'))
predicate_match = predicate == safe_get(a, 'predicate')
edge_label = safe_get(a, 'edge', 'label')
if edge_label is not None:
edge_label = edge_label.replace(' ', '_')
predicate = predicate.replace(' ', '_')
label_match = predicate == edge_label
if object_match and (label_match or predicate_match):
provided_by = safe_get(a, 'edge', 'provided_by')
probability = safe_get(a, 'edge', 'probability')
predicate = safe_get(a, 'predicate')
is_defined_by = safe_get(a, 'api')
endpoint = safe_get(a, 'endpoint')
annotations = []
if probability is not None:
annotations.append(
BeaconStatementAnnotation(tag='probability',
value=probability))
if predicate is not None:
annotations.append(
BeaconStatementAnnotation(tag='predicate',
value=predicate))
if endpoint is not None:
annotations.append(
BeaconStatementAnnotation(tag='endpoint',
value=endpoint))
annotations.append(
BeaconStatementAnnotation(tag='endpoint_input',
value=subject_id))
return BeaconStatementWithDetails(provided_by=provided_by,
is_defined_by=is_defined_by,
annotation=annotations)
return BeaconStatementWithDetails()
def get_predicates(): # noqa: E501
"""get_predicates
Get a list of predicates used in statements issued by the knowledge source # noqa: E501
:rtype: List[BeaconPredicate]
"""
response = requests.get('http://biothings.io/explorer/api/v2/knowledgemap')
predicates = []
if response.ok:
data = response.json()
s = set()
for a in data['associations']:
s.add(safe_get(a, 'predicate'))
for p in s:
if p == p.lower():
predicates.append(BeaconPredicate(edge_label=p))
else:
predicates.append(BeaconPredicate(relation=p))
return predicates
def get_exact_matches_to_concept_list(c):
response = []
for curie in c:
if not isinstance(curie, str) or ':' not in curie:
continue
exact_matches = []
data = crawler.crawl(curie)
for category, associations in data.items():
for a in associations:
if a.get('predicate') == 'EquivalentAssociation':
object_id = safe_get(a, 'object', 'id')
prefix, local_id = object_id.split(':', 1)
if 'name' not in prefix.lower():
exact_matches.append(object_id)
response.append(
ExactMatchResponse(id=curie,
within_domain=data != {},
has_exact_matches=exact_matches))
return response
def find_subject_name(data: dict):
subject_names = []
for category, associations in data.items():
for a in associations:
if a.get('predicate') == 'EquivalentAssociation':
object_id = safe_get(a, 'object', 'id')
object_name = safe_get(a, 'object', 'label')
if object_name != None:
subject_names.append(object_name)
else:
object_prefix, object_local_id = object_id.split(':', 1)
if 'name' in object_prefix.lower(
) or 'symbol' in object_prefix.lower():
subject_names.append(object_local_id)
predicate_longest_under_sixty = lambda n: (len(n) > 60, -len(n))
subject_names.sort(key=predicate_longest_under_sixty)
return next((name for name in subject_names), None)
def get_knowledge_map(): # noqa: E501
"""get_knowledge_map
Get a high level knowledge map of the all the beacons by subject semantic type, predicate and semantic object type # noqa: E501
:rtype: List[BeaconKnowledgeMapStatement]
"""
response = requests.get('http://biothings.io/explorer/api/v2/knowledgemap')
d = defaultdict(lambda: defaultdict(set))
statements = []
if response.ok:
data = response.json()
for a in data['associations']:
subject_category = safe_get(a, 'subject', 'semantic_type')
subject_prefix = safe_get(a, 'subject', 'prefix')
predicate = safe_get(a, 'predicate')
object_category = safe_get(a, 'object', 'semantic_type')
object_prefix = safe_get(a, 'object', 'prefix')
endpoint = safe_get(a, 'endpoint')
k = '{}{}{}'.format(subject_category, predicate, object_category)
d[k]['subject_prefix'].add(subject_prefix)
d[k]['object_prefix'].add(object_prefix)
d[k]['subject_category'] = subject_category
d[k]['object_category'] = object_category
d[k]['predicate'] = predicate
d[k]['endpoint'] = endpoint
for p in d.values():
o = BeaconKnowledgeMapObject(category=p['object_category'],
prefixes=list(p['object_prefix']))
s = BeaconKnowledgeMapSubject(category=p['subject_category'],
prefixes=list(p['subject_prefix']))
if p['predicate'] == p['predicate'].lower():
args = {'edge_label': p['predicate']}
else:
args = {'relation': p['predicate']}
r = BeaconKnowledgeMapPredicate(negated=False, **args)
statements.append(
BeaconKnowledgeMapStatement(subject=s, object=o, predicate=r))
return statements
def get_predicates(): # noqa: E501
"""get_predicates
Get a list of predicates used in statements issued by the knowledge source # noqa: E501
:rtype: List[BeaconPredicate]
"""
data = utils.load_kmap()
predicates = []
s = set()
for a in data['associations']:
s.add(safe_get(a, 'predicate'))
for p in s:
if p.replace('_', ' ') in blm.schema().slots:
edge_label = p.replace(' ', '_')
predicate_id = f'blm:{edge_label}'
relation = edge_label
uri = blm.slot_uri(edge_label)
else:
predicate_id = f'blm:{blm.DEFAULT_EDGE_LABEL}'
edge_label = blm.DEFAULT_EDGE_LABEL
relation = p
uri = blm.slot_uri(blm.DEFAULT_EDGE_LABEL)
try:
description = blm.schema().slots[edge_label.replace(
'_', ' ')].description
except:
description = None
predicates.append(
BeaconPredicate(description=description,
edge_label=edge_label,
relation=relation,
frequency=None))
return predicates
def get_concept_details(conceptId):
data = crawler.crawl(conceptId)
if data == {}:
return BeaconConceptWithDetails()
names, descriptions, xrefs = [], [], []
for category, associations in data.items():
for a in associations:
predicate = a.get('predicate')
if predicate == 'EquivalentAssociation' or predicate == 'HasDescriptionAssociation':
object_id = safe_get(a, 'object', 'id')
prefix, local_id = object_id.split(':', 1)
if 'name' in prefix.lower() or 'symbol' in prefix.lower():
names.append(local_id)
if 'description' in prefix.lower():
descriptions.append(local_id)
if not 'name' in prefix.lower(
) and not 'description' in prefix.lower():
xrefs.append(object_id)
names = list(set(names))
predicate_longest_under_sixty = lambda n: (len(n) > 60, -len(n))
names.sort(key=predicate_longest_under_sixty)
prefix, _ = conceptId.split(':', 1)
c = BeaconConceptWithDetails(id=conceptId,
name=names[0] if len(names) >= 1 else None,
synonyms=names[1:],
exact_matches=xrefs,
categories=[lookup_category(prefix)],
description='; '.join(descriptions))
return c
def get_knowledge_map(): # noqa: E501
"""get_knowledge_map
Get a high level knowledge map of the all the beacons by subject semantic type, predicate and semantic object type # noqa: E501
:rtype: List[BeaconKnowledgeMapStatement]
"""
data = utils.load_kmap()
d = defaultdict(lambda: defaultdict(set))
statements = []
for a in data['associations']:
subject_category = safe_get(a, 'subject', 'semantic_type')
subject_prefix = safe_get(a, 'subject', 'prefix')
predicate = safe_get(a, 'predicate')
object_category = safe_get(a, 'object', 'semantic_type')
object_prefix = safe_get(a, 'object', 'prefix')
endpoint = safe_get(a, 'endpoint')
k = f'{subject_category}{predicate}{object_category}'
d[k]['subject_prefix'].add(subject_prefix)
d[k]['object_prefix'].add(object_prefix)
d[k]['subject_category'] = subject_category
d[k]['object_category'] = object_category
d[k]['predicate'] = predicate
d[k]['endpoint'] = endpoint
for p in d.values():
object_category = p['object_category']
subject_category = p['subject_category']
if object_category not in blm.schema().classes:
object_category = blm.DEFAULT_CATEGORY
if subject_category not in blm.schema().classes:
subject_category = blm.DEFAULT_CATEGORY
if p['predicate'].replace('_', ' ') in blm.schema().slots:
edge_label = p['predicate'].replace(' ', '_')
relation = edge_label
else:
edge_label = blm.DEFAULT_EDGE_LABEL
relation = p['predicate']
o = BeaconKnowledgeMapObject(category=object_category,
prefixes=list(p['object_prefix']))
s = BeaconKnowledgeMapSubject(category=subject_category,
prefixes=list(p['subject_prefix']))
r = BeaconKnowledgeMapPredicate(edge_label=edge_label,
relation=relation,
negated=False)
statements.append(
BeaconKnowledgeMapStatement(subject=s, object=o, predicate=r))
return statements
def get_statements(s=None,
s_keywords=None,
s_categories=None,
edge_label=None,
relation=None,
t=None,
t_keywords=None,
t_categories=None,
offset=None,
size=None):
if s is None:
abort(400,
'Cannot search for statements without providing a subject ID')
statements = []
for subject_id in s:
if ':' not in subject_id:
continue
data = crawler.crawl(subject_id)
if data == {}:
continue
subject_name = find_subject_name(data)
for category, associations in data.items():
if category == 'null':
category = None
for a in associations:
object_id = safe_get(a, 'object', 'id')
if object_id != None and ':' in object_id:
object_prefix, _ = object_id.split(':', 1)
object_prefix = object_prefix.lower()
if 'name' in object_prefix or 'description' in object_prefix:
continue
object_name = safe_get(a, 'object', 'label')
if object_name == None:
secondary_id = safe_get(a, 'object', 'secondary-id')
if secondary_id != None and ':' in secondary_id:
secondary_prefix, symbol = secondary_id.split(':', 1)
object_name = symbol
if 'symbol' in secondary_prefix.lower():
taxonomy = safe_get(a, 'object', 'taxonomy')
if taxonomy != None:
taxonomy = ', '.join(t for t in taxonomy
if ':' not in t)
object_name += ' (taxonomy: {})'.format(
taxonomy)
subject_prefix, _ = subject_id.split(':', 1)
predicate_name = safe_get(a, 'edge', 'label')
if predicate_name == None:
predicate_name = safe_get(a, 'predicate')
if predicate_name == 'EquivalentAssociation':
predicate_name = 'same_as'
if isinstance(predicate_name, list):
if predicate_name != []:
predicate_name = predicate_name[0]
else:
predicate_name = blm.DEFAULT_EDGE_LABEL
predicate_name = predicate_name.replace(' ', '_')
object_id = simplify_curie(object_id)
subject_id = simplify_curie(subject_id)
statements.append(
build_statement(
object_id=object_id,
object_name=object_name,
object_category=category,
subject_id=subject_id,
subject_name=subject_name,
subject_category=utils.lookup_category(subject_prefix),
predicate_id=safe_get(a, 'edge', 'id'),
predicate_name=predicate_name))
is_valid = build_filter(s, s_keywords, s_categories, edge_label, relation,
t, t_keywords, t_categories)
statements = [s for s in statements if is_valid(s)]
statements = remove_duplicates(statements)
if offset is not None:
statements = statements[offset:]
if size is not None:
statements = statements[:size]
return statements
