-
Notifications
You must be signed in to change notification settings - Fork 0
/
DNN.py
225 lines (190 loc) · 8.95 KB
/
DNN.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
import os
import gc
import random
import time
import pandas as pd
from tensorflow import keras
from utils import *
from sklearn.preprocessing import LabelEncoder,MinMaxScaler
from tensorflow.python.keras.callbacks import EarlyStopping
import tensorflow as tf
from tensorflow.python.keras import backend as K
from sklearn.decomposition import PCA
from keras.models import Model
from keras.callbacks import EarlyStopping
from keras.layers import Dense, Dropout, Input, Activation, BatchNormalization
from numpy.random import seed
from keras.backend import clear_session
def init_layer(layer):
session = K.get_session()
weights_initializer = tf.variables_initializer(layer.weights)
session.run(weights_initializer)
def get_embedding(train_d,test_d,get_scaled,n_components):
saver = tf.train.import_meta_graph('./model/KGCN_model_concat64_n7_l2/model_1.ckpt.meta')
from tensorflow.python import pywrap_tensorflow
with tf.Session() as sess:
saver.restore(sess, "./model/KGCN_model_concat64_n7_l2/model_1.ckpt")
model_dir = "./"
checkpoint_path = os.path.join(model_dir,"./model/KGCN_model_concat64_n7_l2/model_1.ckpt")
reader = pywrap_tensorflow.NewCheckpointReader(checkpoint_path)
var_to_shape_map = reader.get_variable_to_shape_map()
for key in var_to_shape_map:
#print("tensor_name: ", key)
if key=='entity_emb_matrix/Adam_1':
entity_emb_matrix=reader.get_tensor(key)
#print(entity_emb_matrix.shape)
#print(test_d[:, 0])
train_drug1 = tf.nn.embedding_lookup(entity_emb_matrix, train_d[:, 0])
train_drug2 = tf.nn.embedding_lookup(entity_emb_matrix, train_d[:, 1])
test_drug1 = tf.nn.embedding_lookup(entity_emb_matrix, test_d[:, 0])
test_drug2 = tf.nn.embedding_lookup(entity_emb_matrix, test_d[:, 1])
with tf.Session() as sess:
train_drug1=train_drug1.eval()
train_drug2=train_drug2.eval()
test_drug1=test_drug1.eval()
test_drug2=test_drug2.eval()
train_feat=np.concatenate([train_drug1,train_drug2],axis=1)
test_feat=np.concatenate([test_drug1,test_drug2],axis=1)
train_feat=mms.fit_transform(train_feat)
test_feat=mms.fit_transform(test_feat)
if get_scaled:
pca = PCA(n_components=n_components)
train_feat = pca.fit_transform(train_feat) #降维
test_feat = pca.transform(test_feat) #归一化
else:
train_feat = train_feat
test_feat = test_feat
return train_feat,test_feat
def get_features(data,drugid_df,use_pro):
drug1_features = pd.merge(data,drugid_df,how='left',left_on='head',right_on='drug_id').iloc[:,4:1029].values
drug2_features = pd.merge(data,drugid_df,how='left',left_on='tail',right_on='drug_id').iloc[:,4:1029].values
if use_pro:
feature = np.concatenate([drug1_features,drug2_features],axis=1)
else:
feature = drug1_features
return feature
def train(i,n_components, use_pro, train_data, test_data,get_scaled):
train_label = train_data[:, 2]
train_label = keras.utils.to_categorical(train_label, 2)
test_label = test_data[:, 2]
test_label = keras.utils.to_categorical(test_label, 2)
#print(type(test_d))
train_feat, test_feat = get_embedding(train_data, test_data,get_scaled,n_components)
train_data = pd.DataFrame(train_data, columns=['head', 'tail', 'label'])
test_data = pd.DataFrame(test_data, columns=['head', 'tail', 'label'])
clear_session()
def DNN():
train_input = Input(shape=(vector_size * 2,), name='Inputlayer')
train_in = Dense(512, activation='relu')(train_input)
train_in = BatchNormalization()(train_in)
train_in = Dropout(droprate)(train_in)
train_in = Dense(256, activation='relu')(train_in)
train_in = BatchNormalization()(train_in)
train_in = Dropout(droprate)(train_in)
train_in = Dense(128, activation='relu')(train_in)
train_in = BatchNormalization()(train_in)
train_in = Dropout(droprate)(train_in)
train_in = Dense(64, activation='relu')(train_in)
train_in = BatchNormalization()(train_in)
train_in = Dropout(droprate)(train_in)
train_in = Dense(event_num)(train_in)
out = Activation('softmax')(train_in)
model = Model(input=train_input, output=out)
# adam = keras.optimizers.Adam(lr=0.001,beta_1=0.9,beta_2=0.999,epsilon=None,decay=0.0)
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy']) # 分类交叉熵
return model
#DNN
print("\n\n------------------------train: %d flod--------------------------------------------------------"%i)
with tf.Graph().as_default():
train_des = get_features(train_data, drugid_df, use_pro)
test_des = get_features(test_data, drugid_df, use_pro)
train_all_feats = np.concatenate([train_feat, train_des], axis=1)
test_all_feats = np.concatenate([test_feat, test_des], axis=1)
dnn = DNN()
for layer in dnn.layers:
init_layer(layer)
print("start train now")
early_stopping = EarlyStopping(monitor='loss', patience=10, verbose=0, mode='auto')
dnn.fit(train_all_feats, train_label, batch_size=1200, epochs=100, validation_data=(test_all_feats, test_label),
callbacks=[early_stopping])
print("--------finish training, test model!!-------------------")
pred = dnn.predict(test_all_feats, batch_size=512)
roc_dnn = roc_auc(test_label[:, 1], pred[:, 1])
pr_dnn = pr_auc(test_label[:, 1], pred[:, 1])
##save model---------------------------------
#savePath = "models_2500/%d_flods/"%i
#os.makedirs(savePath,exist_ok=True)
#dnn.save(savePath+'DNN_2500-%d.h5'%(i))
##----------------------------------------------
print(roc_dnn)
print(pr_dnn)
p = pred[:, [1]]
p[p >= 0.5] = 1
p[p < 0.5] = 0
f1, acc, Pre, Sen, Spe, MCC = scores(test_label[:, 1], p)
train_log = {'agg': 'concat', 'dim': '64', 'AED': '12'}
train_log['test_auc'] = roc_dnn
train_log['test_aupr'] = pr_dnn
train_log['test_acc'] = acc
train_log['test_f1'] = f1
train_log['test_pre'] = Pre
train_log['test_sen'] = Sen
train_log['test_spe'] = Spe
train_log['test_MCC'] = MCC
train_log['timestamp'] = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime())
write_log('./results/DNN_concat64_12AED_per.txt', log=train_log, mode='a')
del dnn
gc.collect()
K.clear_session()
print("-------------------finish train: %d flod--------------------------------------------------------"%i)
return train_log
def cross_validation(K_fold, examples):
subsets = dict()
n_subsets = int(len(examples) / K_fold)
remain = set(range(0, len(examples) - 1))
for i in reversed(range(0, K_fold - 1)):
subsets[i] = random.sample(remain, n_subsets)
remain = remain.difference(subsets[i])
subsets[K_fold - 1] = remain
temp = {'avg_auc': 0.0, 'avg_acc': 0.0, 'avg_f1': 0.0, 'avg_aupr': 0.0, 'avg_pre': 0.0, 'avg_sen': 0.0,
'avg_spe': 0.0, 'avg_MCC': 0.0}
for i in reversed(range(0, K_fold)):
test_data = examples[list(subsets[i])]
train_data = []
for j in range(0, K_fold):
if i != j:
train_data.extend(examples[list(subsets[j])])
train_data = np.array(train_data)
test_data = np.array(test_data)
train_log = train(i,200, True, train_data, test_data,False)
temp['avg_auc'] = temp['avg_auc'] + train_log['test_auc']
temp['avg_acc'] = temp['avg_acc'] + train_log['test_acc']
temp['avg_f1'] = temp['avg_f1'] + train_log['test_f1']
temp['avg_aupr'] = temp['avg_aupr'] + train_log['test_aupr']
temp['avg_pre'] = temp['avg_pre'] + train_log['test_pre']
temp['avg_sen'] = temp['avg_sen'] + train_log['test_sen']
temp['avg_spe'] = temp['avg_spe'] + train_log['test_spe']
temp['avg_MCC'] = temp['avg_MCC'] + train_log['test_MCC']
temp['timestamp'] = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime())
for key in temp:
if key == 'timestamp':
continue
temp[key] = temp[key] / K_fold
write_log('./results/DNN_concat64_12AED_results.txt', temp, 'a')
print(
f'Logging Info - {K_fold} fold result: avg_auc: {temp["avg_auc"]}, avg_acc: {temp["avg_acc"]}, avg_f1: {temp["avg_f1"]}, avg_aupr: {temp["avg_aupr"]}, avg_pre: {temp["avg_pre"]}, avg_sen: {temp["avg_sen"]}, avg_spe: {temp["avg_spe"]}, avg_MCC: {temp["avg_MCC"]}')
seed(1)
mms = MinMaxScaler(feature_range=(0,1))
FIRST=True
event_num = 2
droprate = 0.5
vector_size = 1088
drug_id = pd.read_csv('./data/drug_index.csv',delimiter=',', header=None)
drug_id.columns = ['drug','drug_id']
drug_id = drug_id['drug_id']
#descriptors preparation
drug_feats = np.loadtxt('./data/MorganFinger.txt',delimiter=' ')
drugid_df = pd.concat([drug_id,pd.DataFrame(drug_feats)],axis=1)
DDI_softmax='./data/binary_DDI.txt'
drug1_set, drug2_set, DDI = readRecData(DDI_softmax)
cross_validation(10, np.array(DDI))