def get_IC():
cur = db.cursor()
# check for exact match
query = """SELECT distinct term_id, info_content
FROM SSM"""
cur.execute(query)
res = cur.fetchall()
return zip(*res)
def check_dist_between(cid1, cid2):
cur = db.cursor()
query = """SELECT distance
FROM graph_path
WHERE term1_id = %s and term2_id = %s""" % (cid1, cid2)
cur.execute(query)
res = cur.fetchone()
if res is None:
dist = -1
else:
dist = int(res[0])
return dist
def get_description(id):
cur = db.cursor()
query = """SELECT term_definition
FROM term_definition
WHERE term_id = %s""" % id
cur.execute(query)
res = cur.fetchone()
if res is not None:
return res[0]
else:
return "NA"
def load_synonyms():
syns = []
cur = db.cursor()
query = """SELECT id, name
FROM term """
cur.execute(query)
ids = cur.fetchall()
for i in ids:
print "getting synonyms for" + i[1].lower() + '(' + str(i[0]) + ')',
synset = set()
synset.add(i[1].lower())
query = """SELECT term_synonym
FROM term_synonym
WHERE term_id = %s""" % i[0]
cur.execute(query)
names = cur.fetchall()
print len(names)
for name in names:
#print name[0],
synset.add(name[0].lower())
syns.append(synset)
pickle.dump(syns, open("data/hpo_synonyms.pickle", 'wb'))
print "done"
def find_hpo_term(term, adjust=0):
''' returns tuple (hpoID, hpoTerm, score)
if resolution fails, return ('0', 'null', 0.0)
'''
# print "TERM", term
term = MySQLdb.escape_string(term)
# adjust - adjust the final score
match = ()
cur = db.cursor()
# check for exact match
query = """SELECT distinct id, name
FROM term a
WHERE name =%s and LENGTH(a.name)>0;"""
# print "QUERY", query
cur.execute(query, (term, ))
res = cur.fetchone()
if res is not None:
# print "1"
score = 1.0 + adjust
match = (str(res[0]), res[1], score)
else:
# synonyms
cur.execute(
"""SELECT a.term_id, a.term_synonym, b.name
FROM term_synonym a, term b
WHERE a.term_synonym=%s
and b.id=a.term_id
and LENGTH(a.term_synonym)>0""", (term, ))
res = cur.fetchone()
if res is not None:
# print "2"
score = 0.8 + adjust
match = (str(res[0]), res[2], score)
else:
# plural - tb pode ser recursivo
if len(term) > 0 and term[-1] == 's':
match = find_hpo_term(term[:-1], -0.1)
#######################################################
##HPO has no descriptor table and no ec (in same table)
# terms = '("' + '","'.join(term.split(" ")) + '")'
# ## JOins several tables and selects dictint words from term list.
# query = """SELECT ((sum(d.ic)/ec)-0.1) as score, e.name, c.term_id, c.id,
# group_concat(d.word separator ','), count(d.id), c.descriptor_type
# FROM term e JOIN descriptor3 c ON(c.term_id=e.id) JOIN word2term3 b ON (b.descriptor_id=c.id)
# JOIN word3 d ON (d.id=b.word_id) JOIN SSM_TermDesc f ON (e.id=f.term_id)
# WHERE b.word_id IN (
# SELECT distinct id
# FROM word3
# WHERE word in %s)
# GROUP by c.id
# ORDER by score desc
# LIMIT 3;""" % (terms,)
# # print "QUERY3", query, adjust
# cur.execute(query)
# res = cur.fetchone()
# if res is not None:
# # print "3"
# match = (str(res[3]), res[1], float(res[0]))
# # print term, match
######################################################
if not match or match[2] < 0.0:
match = ('0', 'null', 0.0)
return match