#!/usr/bin/python
from http_wrapper import HttpWrapper
from urllib.parse import quote
import sys
START_S_ID = None
if len(sys.argv) > 1:
START_S_ID = sys.argv[1]
hgnc_search_uri = 'http://rest.genenames.org'
db_uri = 'http://e3cb0988.ngrok.io'
# Get all preprocessed MeSH terms in DB.
db_http = HttpWrapper(db_uri)
all_processed = db_http.request(
'/sql',
'GET',
{
'query': 'SELECT * FROM LUNG_PROCESSED WHERE S_ID = 28',
},
'',
)
# Search using P_NAME from processed rows.
# S_ID and HGNC Map.
# Key: S_ID
# Value: HGNC doc max score
sid_hgnc_map = dict()
password='',
db='mesh',
charset='utf8')
elapsed_millis = get_current_millis()
# Get all genes which does not have MESH_NAME.
all_genes = None
with db.cursor(pymysql.cursors.DictCursor) as cursor:
query = 'SELECT * FROM PROSTATE_GENES WHERE MESH_NAME IS NULL ORDER BY S_ID' if START_S_ID is None \
else 'SELECT * FROM PROSTATE_GENES WHERE S_ID > %s AND MESH_NAME IS NULL ORDER BY S_ID' % (START_S_ID)
cursor.execute(query)
all_genes = cursor.fetchall()
print('Find all genes time:',
get_elapsed_seconds(get_current_millis(), elapsed_millis))
hgnc_http = HttpWrapper('http://rest.genenames.org')
def get_max_score_doc(result_docs):
return max(result_docs, key=lambda doc: doc['score'])
# Fulfill gene's MESH_NAME.
for gene in all_genes:
elapsed_millis = get_current_millis()
processeds = None
with db.cursor(pymysql.cursors.DictCursor) as cursor:
cursor.execute('SELECT * FROM PROSTATE_PROCESSED where S_ID = %s',
(gene['S_ID'], ))
processeds = cursor.fetchall()
print('Get processed time:',
all_qualifiers = None
with db.cursor(pymysql.cursors.DictCursor) as cursor:
cursor.execute('SELECT * FROM MESH_QUALIFIER')
all_qualifiers = cursor.fetchall()
all_descriptors = None
with db.cursor(pymysql.cursors.DictCursor) as cursor:
cursor.execute('SELECT * FROM MESH_DESCRIPTOR')
all_descriptors = cursor.fetchall()
all_supplementals = None
with db.cursor(pymysql.cursors.DictCursor) as cursor:
cursor.execute('SELECT * FROM MESH_SUPPLEMENTAL')
all_supplementals = cursor.fetchall()
print('Find all mesh terms time:',
get_elapsed_seconds(get_current_millis(), elapsed_millis))
hgnc_http = HttpWrapper(hgnc_search_uri)
def get_max_score_doc(result_docs):
return max(result_docs, key=lambda doc: doc['score'])
def check_is_family(mesh_term):
qualifiers = list(
filter(lambda qualifier: qualifier['NAME'] == mesh_term,
all_qualifiers))
descriptors = list(
filter(lambda descriptor: descriptor['NAME'] == mesh_term,
all_descriptors))
supplementals = list(
filter(lambda supplemental: supplemental['NAME'] == mesh_term,
all_descriptors = cursor.fetchall()
all_supplementals = None
with db.cursor(pymysql.cursors.DictCursor) as cursor:
cursor.execute('SELECT * FROM MESH_SUPPLEMENTAL')
all_supplementals = cursor.fetchall()
print('Find all mesh terms time:',
get_elapsed_seconds(get_current_millis(), elapsed_millis))
# Search using P_NAME from processed rows.
# S_ID and HGNC Map.
# Key: S_ID
# Value: HGNC doc max score
sid_hgnc_map = dict()
hgnc_http = HttpWrapper(hgnc_search_uri)
def get_max_score_doc(result_docs):
return max(result_docs, key=lambda doc: doc['score'])
def check_is_family(mesh_term):
qualifiers = list(
filter(lambda qualifier: qualifier['NAME'] == mesh_term,
all_qualifiers))
descriptors = list(
filter(lambda descriptor: descriptor['NAME'] == mesh_term,
all_descriptors))
supplementals = list(
filter(lambda supplemental: supplemental['NAME'] == mesh_term,