import best
q = best.BESTQuery("breast cancer", "gene", noAbsTxt=False)
r = best.getRelevantBioEntities(q)
print(r[0])
def best_search(self, query, kcw_path=None):
t0 = time()
# Type filter
ent_types = [
"gene", "drug", "chemical compound", "target", "disease",
"toxin", "transcription factor", "mirna", "pathway", "mutation"
]
query_type = "All Entity Type"
for ent_type in ent_types:
if ent_type in query:
query_type = ent_type
break
# Stopwords and filtering for BEST queries
if not os.path.exists(os.path.join(os.path.expanduser('~'), 'nltk_data')):
nltk.download('punkt')
nltk.download('stopwords')
stop_words = set(stopwords.words('english') + ['?'] + ['Why', 'What', 'How', 'Where', 'When', 'Who'])
entity_set = [
'COVID-19', 'SARS-CoV-2', 'hypertension', 'diabetes', 'heart', 'disease', 'obese', 'death',
'HCoV-19', 'HCoV', 'coronavirus', 'symptoms', 'incubation', 'periods', 'period', 'quarantine',
'asymptomatic', 'transmissions', 'fecal', 'excretion', 'decline', 'Wuhan', 'mortality',
'patients', 'stay', 'reproduction', 'risk', 'factor', 'factors', 'pregnancy', 'interval', 'absent',
'reported', 'length', 'diagnosed', 'United', 'States', 'isolated', 'CDC', 'WHO', 'vaccine',
'negative', 'animals', 'airbone', 'spread', 'blood', 'sanitizer', 'controlled', 'illness', 'friends',
]
query_tokens = word_tokenize(query)
new_query = ''
for idx, query_token in enumerate(query_tokens):
if query_token not in stop_words:
if query_token in entity_set:
new_query += query_token + ' '
# Get BEST result
q = best.BESTQuery(new_query, noAbsTxt=False, filterObjectName=query_type)
r = best.getRelevantBioEntities(q)
# No result
if len(r) == 1 and r[0]['rank'] == 0 and len(r[0].keys()) == 1:
t1 = time()
return {'ret': [], 'time': int(1000 * (t1 - t0))}
parsed_result = {
'context': '',
'title': '',
'doc_idx': None,
'start_idx': 0,
'end_idx': 0,
'score': 0,
'metadata': {
'pubmed_id': ''
},
'answer': ''
}
outs = []
for r_idx, r_ in enumerate(r):
parsed_result['context'] = r_['abstracts'][0]
parsed_result['score'] = r_['score']
parsed_result['answer'] = r_['entityName']
parsed_result['metadata'] = self.get_doc_meta(r_['PMIDs'][0])
if len(parsed_result['metadata']) == 0:
parsed_result['metadata']['pubmed_id'] = int(r_['PMIDs'][0])
outs.append(copy.deepcopy(parsed_result))
t1 = time()
return {'ret': outs, 'time': int(1000 * (t1 - t0))}
import best
bestQuery = best.BESTQuery({
"keywordA": ["cancer"],
"keywordB": ["BRCA1", "EGFR"],
"filterObjectName": "",
"topN": 30
})
relevantEntities = best.getRelevantBioEntities(bestQuery)
print(relevantEntities)