def normalize(self):
term = MySQLdb.escape_string(self.text)
# adjust - adjust the final score
match = ()
cur = db.cursor()
# synonym
query = """SELECT DISTINCT t.acc, t.name, s.term_synonym
FROM term t, term_synonym s
WHERE s.term_synonym LIKE %s and s.term_id = t.id
ORDER BY t.ic ASC
LIMIT 1;""" # or DESC
# print "QUERY", query
cur.execute(query, ("%" + term + "%", ))
res = cur.fetchone()
if res is not None:
print res
else:
query = """SELECT DISTINCT t.acc, t.name, p.name
FROM term t, prot p, prot_GOA_BP a
WHERE p.name LIKE %s and p.id = a.prot_id and a.term_id = t.id
ORDER BY t.ic ASC
LIMIT 1;""" # or DESC
cur.execute(query, (term, ))
res = cur.fetchone()
print res
def simui_hindex_go(id1, id2, h=4):
cur = dbwebgo.cursor()
query = """SELECT (
SELECT COUNT(y.id)
FROM (
SELECT DISTINCT t3.id
FROM graph_path p1, graph_path p2, term t1, term t2, term t3
WHERE t1.acc = %s AND t2.acc = %s AND
t1.id = p1.term2_id AND t2.id = p2.term2_id
AND p1.term1_id=p2.term1_id
AND p1.term1_id=t3.id AND t3.hindex >= %s)
AS y )
/(
SELECT COUNT(x.id)
FROM (
SELECT t2.id
FROM graph_path p1, term t1, term t2
WHERE t1.acc = %s AND t1.id = p1.term2_id
AND p1.term1_id = t2.id AND t2.hindex >= %s
UNION
SELECT t4.id
FROM graph_path p2, term t3, term t4
WHERE t3.acc = %s AND t3.id = p2.term2_id
AND p2.term1_id = t4.id AND t4.hindex >= %s)
AS x )"""
cur.execute(query, (id1,id2,h,id1,h,id2,h))
res = cur.fetchone()[0]
if res is None:
res = '0'
return float(res)
def simgic_hindex_go(id1, id2, h=4):
cur = dbwebgo.cursor()
cur.execute("""SELECT (
SELECT SUM(y.ic)
FROM (
SELECT DISTINCT f.id, f.ic
FROM graph_path p1, graph_path p2, term f, term f1, term f2
WHERE p1.term2_id = f1.id AND f1.acc = %s
AND p2.term2_id = f2.id AND f2.acc = %s
AND p1.term1_id=p2.term1_id
AND p1.term1_id=f.id
AND f.hindex >= %s)
AS y )
/(
SELECT SUM(x.ic)
FROM (
SELECT f1.id, f1.ic
FROM graph_path p1, term f1, term f3
WHERE p1.term2_id = f3.id and f3.acc = %s
AND p1.term1_id = f1.id
AND f1.hindex >= %s
UNION
SELECT f2.id, f2.ic
FROM graph_path p2, term f2, term f4
WHERE p2.term2_id = f4.id and f4.acc = %s
AND p2.term1_id = f2.id
AND f2.hindex >= %s)
AS x )""", (id1,id2,h,id1,h,id2,h))
res = cur.fetchone()[0]
if res is None:
res = '0'
return float(res)
def resnik_go(id1, id2):
cur = dbwebgo.cursor()
query = """SELECT MAX(t3.ic)
FROM graph_path p1, graph_path p2, term t1, term t2, term t3
WHERE t1.acc = %s AND t2.acc = %s AND t1.id = p1.term2_id AND t2.id = p2.term2_id
AND p1.term1_id = p2.term1_id
AND p1.term1_id = t3.id;"""
cur.execute(query, (id1, id2))
#r = cur.store_result()
res = cur.fetchone()[0]
#print id1, id2, res
if res is None:
res = '0'
return float(res)
def get_best_go(self):
cur = db.cursor()
# synonym
query = """SELECT DISTINCT t.acc, t.name, t.ic
FROM term t
WHERE t.acc IN (%s)
ORDER BY t.ic ASC
LIMIT 1;""" # or DESC
# print "QUERY", query
format_strings = ','.join(['%s'] * len(self.go_ids))
cur.execute(query % format_strings, (self.go_ids))
res = cur.fetchone()
if res is not None:
# print self.text, res[1:]
logging.info("best GO for {} ({}): {}".format(
self.text, len(self.go_ids), " ".join([str(r) for r in res])))
self.best_go = res[0]
else:
logging.info("NO GO")
self.best_go = ""
def get_best_go(self):
cur = db.cursor()
# synonym
query = """SELECT DISTINCT t.acc, t.name, t.ic
FROM term t
WHERE t.acc IN (%s)
ORDER BY t.ic ASC
LIMIT 1;""" # or DESC
# print "QUERY", query
format_strings = ','.join(['%s'] * len(self.go_ids))
cur.execute(query % format_strings, (self.go_ids))
res = cur.fetchone()
if res is not None:
# print self.text, res[1:]
logging.info("best GO for {} ({}): {}".format(self.text, len(self.go_ids), " ".join([str(r) for r in res])))
self.best_go = res[0]
else:
logging.info("NO GO")
self.best_go = ""
