def run_analysis(self):
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 #number of random miRNA-gene pairs with common go
all_tfs = []
all_mirnas = []
for pmid in self.tfs:
all_tfs = []
all_mirnas = []
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 # number of random miRNA-gene pairs with common go
for tf in self.tfs[pmid]:
all_tfs.append(tf)
mirna = None
for mirna_eid in tf.targets:
for m in self.mirnas[pmid]:
if m.eid == mirna_eid[0]:
mirna = m
break
all_mirnas.append(mirna)
# common_gos = set(tf.go_ids).intersection(set(mirna.go_ids))
# if len(common_gos) > 0:
# print "{}->{} common gos:{}".format(tf.text, mirna.text, " ".join(common_gos))
# correct_count += 1
if len(all_tfs) > 1 and len(all_mirnas) > 1:
for i in range(0, 10):
random_tf = random.choice(all_tfs)
random_mirna = random.choice(all_mirnas)
common_gos = set(random_tf.go_ids).intersection(
set(random_mirna.go_ids))
if (random_mirna.eid, "miRNA-gene") in random_tf.targets:
#if len(common_gos) > 0:
if random_mirna.best_go.startswith(
"GO:") and random_tf.best_go.startswith("GO"):
ss = ssm.simui_go(random_mirna.best_go,
random_tf.best_go)
#print "correct:", ss
correct_count += ss
else:
correct_count += 1
else:
#if len(common_gos) > 0:
if random_mirna.best_go.startswith(
"GO:") and random_tf.best_go.startswith("GO"):
ss = ssm.simui_go(random_mirna.best_go,
random_tf.best_go)
print "incorrect:", ss
incorrect_count += ss
else:
incorrect_count += 1
print "{}-{} ({} mirnas, {} tfs".format(
correct_count, incorrect_count, len(all_mirnas),
len(all_tfs))
def run_analysis(self):
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 #number of random miRNA-gene pairs with common go
all_tfs = []
all_mirnas = []
for pmid in self.tfs:
all_tfs = []
all_mirnas = []
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 # number of random miRNA-gene pairs with common go
for tf in self.tfs[pmid]:
all_tfs.append(tf)
mirna = None
for mirna_eid in tf.targets:
for m in self.mirnas[pmid]:
if m.eid == mirna_eid[0]:
mirna = m
break
all_mirnas.append(mirna)
# common_gos = set(tf.go_ids).intersection(set(mirna.go_ids))
# if len(common_gos) > 0:
# print "{}->{} common gos:{}".format(tf.text, mirna.text, " ".join(common_gos))
# correct_count += 1
if len(all_tfs) > 1 and len(all_mirnas) > 1:
for i in range(0, 10):
random_tf = random.choice(all_tfs)
random_mirna = random.choice(all_mirnas)
common_gos = set(random_tf.go_ids).intersection(set(random_mirna.go_ids))
if (random_mirna.eid, "miRNA-gene") in random_tf.targets:
#if len(common_gos) > 0:
if random_mirna.best_go.startswith("GO:") and random_tf.best_go.startswith("GO"):
ss = ssm.simui_go(random_mirna.best_go, random_tf.best_go)
#print "correct:", ss
correct_count += ss
else:
correct_count += 1
else:
#if len(common_gos) > 0:
if random_mirna.best_go.startswith("GO:") and random_tf.best_go.startswith("GO"):
ss = ssm.simui_go(random_mirna.best_go, random_tf.best_go)
print "incorrect:", ss
incorrect_count += ss
else:
incorrect_count += 1
print "{}-{} ({} mirnas, {} tfs".format(correct_count, incorrect_count, len(all_mirnas), len(all_tfs))
def annotate_corpus_relations(corpus, model, corpuspath):
logging.info("getting relations...")
# entities, relations = load_gold_relations(reltype)
logging.info("finding relations...")
# print entities.keys()[:20]
for did in corpus.documents:
for sentence in corpus.documents[did].sentences:
sentences_mirnas = []
sentence_tfs = []
#print sentence.entities.elist
for entity in sentence.entities.elist[model]:
if entity.type == "mirna":
sentences_mirnas.append(entity)
elif entity.type == "protein":
sentence_tfs.append(entity)
for mirna in sentences_mirnas:
for tf in sentence_tfs:
ss = ssm.simui_go(mirna.best_go, tf.best_go)
if ss > 0:
print ss, mirna.text, tf.text, mirna.best_go, tf.best_go
print "saving corpus..."
corpus.save(corpuspath)
def annotate_corpus_relations(corpus, model, corpuspath):
logging.info("getting relations...")
# entities, relations = load_gold_relations(reltype)
logging.info("finding relations...")
# print entities.keys()[:20]
for did in corpus.documents:
for sentence in corpus.documents[did].sentences:
sentences_mirnas = []
sentence_tfs = []
#print sentence.entities.elist
for entity in sentence.entities.elist[model]:
if entity.type == "mirna":
sentences_mirnas.append(entity)
elif entity.type == "protein":
sentence_tfs.append(entity)
for mirna in sentences_mirnas:
for tf in sentence_tfs:
ss = ssm.simui_go(mirna.best_go, tf.best_go)
if ss > 0:
print ss, mirna.text, tf.text, mirna.best_go, tf.best_go
print "saving corpus..."
corpus.save(corpuspath)
def run_ss_analysis(self, pairtype):
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 # number of random miRNA-gene pairs with common go
all_tfs = []
all_mirnas = []
diff_count = []
nexp = 10
for did in self.documents:
#for sentence in self.documents[did].sentences:
all_tfs = []
all_mirnas = []
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 # number of random miRNA-gene pairs with common go
all_mirnas = self.documents[did].get_entities('goldstandard_mirna')
all_tfs = self.documents[did].get_entities('goldstandard_protein')
#count true relations
if len(all_mirnas) > 0 and len(all_tfs) > 0:
# for entity in sentence.entities.elist["goldstandard"]:
# if entity.type == "protein":
# all_tfs.append(entity)
# elif entity.type == "mirna":
# all_mirnas.append(entity)
# correct_count += 1
#if len(all_tfs) > 1 and len(all_mirnas) > 1:
# print sentence.sid
correct = 0
incorrect = 0
while correct < nexp and incorrect < nexp:
random_tf = random.choice(all_tfs)
random_mirna = random.choice(all_mirnas)
# print dir(random_mirna)
# common_gos = set(random_tf.go_ids).intersection(set(random_mirna.go_ids))
if correct < nexp and (random_tf.eid,
pairtype) in random_mirna.targets:
# if len(common_gos) > 0:
# if random_mirna.best_go.startswith("GO:") and random_tf.best_go.startswith("GO"):
# print random_mirna.best_go, random_tf.best_go
max_ss = []
for mirnago in random_mirna.go_ids:
for mirnatf in random_tf.go_ids:
ss = ssm.simui_go(mirnago, mirnatf)
max_ss.append(ss)
#if max_ss > ss:
# max_ss = ss
# ss = ssm.simui_go(random_mirna.best_go, random_tf.best_go)
# print "correct:", ss
if len(max_ss) > 0:
correct_count += sum(max_ss) * 1.0 / len(max_ss)
# correct_count += len(common_gos)
correct += 1
elif incorrect < nexp:
# if len(common_gos) > 0:
# if random_mirna.best_go.startswith("GO:") and random_tf.best_go.startswith("GO"):
max_ss = []
for mirnago in random_mirna.go_ids:
for mirnatf in random_tf.go_ids:
# ss = ssm.simui_hindex_go(mirnago, mirnatf, h=0)
ss = ssm.simui_go(mirnago, mirnatf)
max_ss.append(ss)
#if max_ss > ss:
# max_ss = ss
# ss = ssm.simui_go(random_mirna.best_go, random_tf.best_go)
# print "correct:", ss
if len(max_ss) > 0:
incorrect_count += sum(max_ss) * 1.0 / len(max_ss)
# incorrect_count += len(common_gos)
incorrect += 1
if correct_count != 0 and incorrect_count != 0:
print "{}={}-{} ({} mirnas, {} tfs)".format(
correct_count / nexp - incorrect_count / nexp,
correct_count / nexp, incorrect_count / nexp,
len(all_mirnas), len(all_tfs))
diff_count.append(correct_count / nexp -
incorrect_count / nexp)
print sum(diff_count) * 1.0 / len(diff_count)
def run_ss_analysis(self, pairtype):
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 # number of random miRNA-gene pairs with common go
all_tfs = []
all_mirnas = []
diff_count = []
nexp = 10
for did in self.documents:
#for sentence in self.documents[did].sentences:
all_tfs = []
all_mirnas = []
correct_count = 0 # numver of real miRNA-gene pairs with common gos
incorrect_count = 0 # number of random miRNA-gene pairs with common go
all_mirnas = self.documents[did].get_entities('goldstandard_mirna')
all_tfs = self.documents[did].get_entities('goldstandard_protein')
#count true relations
if len(all_mirnas) > 0 and len(all_tfs) > 0:
# for entity in sentence.entities.elist["goldstandard"]:
# if entity.type == "protein":
# all_tfs.append(entity)
# elif entity.type == "mirna":
# all_mirnas.append(entity)
# correct_count += 1
#if len(all_tfs) > 1 and len(all_mirnas) > 1:
# print sentence.sid
correct = 0
incorrect = 0
while correct < nexp and incorrect < nexp:
random_tf = random.choice(all_tfs)
random_mirna = random.choice(all_mirnas)
# print dir(random_mirna)
# common_gos = set(random_tf.go_ids).intersection(set(random_mirna.go_ids))
if correct < nexp and (random_tf.eid, pairtype) in random_mirna.targets:
# if len(common_gos) > 0:
# if random_mirna.best_go.startswith("GO:") and random_tf.best_go.startswith("GO"):
# print random_mirna.best_go, random_tf.best_go
max_ss = []
for mirnago in random_mirna.go_ids:
for mirnatf in random_tf.go_ids:
ss = ssm.simui_go(mirnago, mirnatf)
max_ss.append(ss)
#if max_ss > ss:
# max_ss = ss
# ss = ssm.simui_go(random_mirna.best_go, random_tf.best_go)
# print "correct:", ss
if len(max_ss) > 0:
correct_count += sum(max_ss)*1.0/len(max_ss)
# correct_count += len(common_gos)
correct += 1
elif incorrect < nexp:
# if len(common_gos) > 0:
# if random_mirna.best_go.startswith("GO:") and random_tf.best_go.startswith("GO"):
max_ss = []
for mirnago in random_mirna.go_ids:
for mirnatf in random_tf.go_ids:
# ss = ssm.simui_hindex_go(mirnago, mirnatf, h=0)
ss = ssm.simui_go(mirnago, mirnatf)
max_ss.append(ss)
#if max_ss > ss:
# max_ss = ss
# ss = ssm.simui_go(random_mirna.best_go, random_tf.best_go)
# print "correct:", ss
if len(max_ss) > 0:
incorrect_count += sum(max_ss)*1.0/len(max_ss)
# incorrect_count += len(common_gos)
incorrect += 1
if correct_count != 0 and incorrect_count != 0:
print "{}={}-{} ({} mirnas, {} tfs)".format(correct_count / nexp - incorrect_count / nexp,
correct_count / nexp, incorrect_count / nexp,
len(all_mirnas), len(all_tfs))
diff_count.append(correct_count / nexp - incorrect_count / nexp)
print sum(diff_count) * 1.0 / len(diff_count)