-
Notifications
You must be signed in to change notification settings - Fork 0
/
sample_expansion.py
256 lines (210 loc) · 11.4 KB
/
sample_expansion.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
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
# -*- coding: utf-8 -*-
"""
Created on Wed Apr 7 22:34:22 2021
@author: ASUS
"""
#######利用找到的motif进行样本扩充,以提高模型准确度
#1.导入数据和模型
#2.导入motif数据并处理
##2.1提取长度小于20的motif
##2.2通过滑动motif在原序列中的位置扩增正面样本
##2.2通过随机化motif两端的无关序列扩增正面样本
#3.将利用扩增的样本和新的反面样本继续训练模型
#4.将继续训练的模型在test数据集上进行预测
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
import tensorflow.keras.layers as layers
class sample_expansion:
def __init__(self,cell_line):
model_path="D:/workspace of spyder/毕业设计/my project data/model_file/Binary_classification_"+cell_line+".h5"
self.model=tf.keras.models.load_model(model_path)
self.model_retrained=None
self.p_file=cell_line
if cell_line=="Dnd41": #DNd细胞系的文件夹命名出了点问题
self.p_file="DND41"
self.x_test=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/x_test_"+cell_line+".npy")
self.y_test=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/y_test_"+cell_line+".npy")
self.x_validate=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/x_validate_"+cell_line+".npy")
self.y_validate=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/y_validate_"+cell_line+".npy")
self.motifs_train_x=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/clear_positive_x_train_motifs_"+self.p_file+".npy")
self.expanded_samples_slide=None
self.expanded_samples_padding=None
self.evaluation_loaded_model=self.model.evaluate(self.x_test,self.y_test)
self.x_train=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/x_train_"+self.p_file+".npy")
self.model_retrained=None
self.background=np.load("D:/workspace of spyder/毕业设计/my project data/datafile/"+self.p_file+"/"+self.p_file+"_background.npy")
#通过滑动motif进行样本扩增
## #先检查motif的长度,选择长度小于等于16的motif
#产生连个随机数r1,r2 ,r1介于0到p之间,r2介于0到99-q之间表示向左移动r1或向右移动r2
#根据motif是在train数据集中的位置进行移动,将新的样本加入到new_samples中
#将new_samples赋值给self.expanded_samples
def expand_sample_slide(self,times):
new_samples=[]
for j in range(times):
for i in range(len(self.motifs_train_x)):
left=self.motifs_train_x[i][0]
right=self.motifs_train_x[i][1]
position=self.motifs_train_x[i][2]
if (right-left)+1<=16 and position<len(self.x_train):
position=self.motifs_train_x[i][2]
sample=np.zeros((len(self.x_test[0]),4))
r1=np.random.randint(left)
r2=np.random.randint(len(self.x_test[0])-1-right)
sample[left-r1:right-r1+1,:]=self.x_train[position][left:right+1,:] #motif段
sample[0:left-r1,:]=self.x_train[position][0:left-r1,:] #最左段
sample[right:len(sample),:]=self.x_train[position][right:len(sample),:]#最右段
sample[right-r1+1:right,:]=self.x_train[position][left-r1+1:left,:] #填补段
new_samples.append(sample)
sample[left+r2:right+r2+1,:]=self.x_train[position][left:right+1,:] #motif段
sample[0:left,:]=self.x_train[position][0:left,:] #最左段
sample[right+r2:len(sample),:]=self.x_train[position][right+r2:len(sample),:]#最右段
sample[left:left+r2,:]=self.x_train[position][right:right+r2,:] #填补段
new_samples.append(sample)
new_samples=np.array(new_samples)
self.expanded_samples_slide=new_samples
return new_samples
##通过补充全A,全C全T全G序列来进行样本扩充
def expand_sample_padding(self):
new_samples=[]
for i in range(len(self.motifs_train_x)):
left=self.motifs_train_x[i][0]
right=self.motifs_train_x[i][1]
position=self.motifs_train_x[i][2]
sample=np.zeros((len(self.x_train[0]),4))
sample[:,0]=1#全部填充为A
sample[left:right,:]=self.x_train[position][left:right,:] #motif段
new_samples.append(list(sample))
sample=np.zeros((len(self.x_train[0]),4))
sample[:,1]=1#全部填充为T
sample[left:right,:]=self.x_train[position][left:right,:] #motif段
new_samples.append(list(sample))
sample=np.zeros((len(self.x_train[0]),4))
sample[:,2]=1#全部填充为C
sample[left:right,:]=self.x_train[position][left:right,:] #motif段
new_samples.append(list(sample))
sample=np.zeros((len(self.x_train[0]),4))
sample[:,3]=1#全部填充为G
sample[left:right,:]=self.x_train[position][left:right,:] #motif段
new_samples.append(list(sample))
new_samples=np.array(new_samples)
self.expanded_samples_padding=new_samples
return new_samples
def retrain_model(self):
if len(self.expanded_samples_slide)==0:
print("no expanded samples")
return 0
x_background=self.background[0:len(self.expanded_samples_slide)+len(self.x_train)]
x=np.append(self.expanded_samples_slide,self.x_train,axis=0)
print(len(x))
x=np.append(x,x_background,axis=0)
y_positive=np.ones(len(self.expanded_samples_slide)+len(self.x_train))
y_background=np.zeros(len(x_background))
y=np.append(y_positive,y_background)
index=[]
for i in range(len(x)):
index.append(i)
np.random.shuffle(index)
x=x[index]
y=y[index]
detctor_length=24
num_detector=32
num_hidden_unit=32
from tensorflow.keras import regularizers
weight_decay = 0.005
# kernel_regularizer=regularizers.l2(weight_decay)
##2.建立模型
model=tf.keras.Sequential()
#model.add(layers.Conv1D(num_detector,detctor_length,input_shape=(train_x.shape[1:]),activation='relu'))
model.add(layers.Conv1D(num_detector,detctor_length,input_shape=(x.shape[1:]),activation='relu',kernel_regularizer=regularizers.l2(weight_decay)))
model.add(layers.GlobalMaxPool1D())
#model.add(layers.Dense(num_hidden_unit,activation='relu'))
model.add(layers.Dense(num_hidden_unit,activation='relu',kernel_regularizer=regularizers.l2(weight_decay)))
model.add(layers.Dropout(0.5))
model.add(layers.Dense(1,activation='sigmoid'))
model.compile(optimizer=tf.keras.optimizers.RMSprop(),
loss='binary_crossentropy',
metrics=['acc'])
##4.模型训练
history = model.fit(x,y,
epochs=25,
batch_size=256,
validation_data=(self.x_validate,self.y_validate))
plt.plot(history.epoch,history.history.get('acc'),label='acc')
plt.plot(history.epoch,history.history.get('val_acc'),label='val_acc')
plt.legend()
self.model_retrained=model
model.save("D:/workspace of spyder/毕业设计/my project data/model_file/retrained_Binary_classification_"+self.p_file+".h5")
## 产生全ATCG的背景序列
def generate_background(self,num):
new_samples=[]
for i in range(num):
if i%4==0:
sample=np.zeros((100,4))
sample[:,0]=1#全部填充为A
new_samples.append(list(sample))
if i%4==1:
sample=np.zeros((100,4))
sample[:,1]=1#全部填充为T
new_samples.append(list(sample))
if i%4==2:
sample=np.zeros((100,4))
sample[:,2]=1#全部填充为C
new_samples.append(list(sample))
if i%4==3:
sample=np.zeros((100,4))
sample[:,3]=1#全部填充为G
new_samples.append(list(sample))
new_samples=np.array(new_samples)
return new_samples
def retrain_model2(self):
if len(self.expanded_samples_padding)==0:
print("no expanded samples")
return 0
x_background=self.background[0:len(self.expanded_samples_padding)+len(self.x_train)*2]
temp=self.generate_background(len(self.expanded_samples_padding)+len(self.x_train))
x_background=np.append(x_background,temp,axis=0)
x=np.append(self.expanded_samples_padding,self.x_train,axis=0)
x=np.append(x,x_background,axis=0)
y_positive=np.ones(len(self.expanded_samples_padding)+len(self.x_train))
y_background=np.zeros(len(x_background))
y=np.append(y_positive,y_background)
index=[]
for i in range(len(x)):
index.append(i)
np.random.shuffle(index)
x=x[index]
y=y[index]
detctor_length=24
num_detector=32
num_hidden_unit=32
from tensorflow.keras import regularizers
weight_decay = 0.005
# kernel_regularizer=regularizers.l2(weight_decay)
##2.建立模型
model=tf.keras.Sequential()
#model.add(layers.Conv1D(num_detector,detctor_length,input_shape=(train_x.shape[1:]),activation='relu'))
model.add(layers.Conv1D(num_detector,detctor_length,input_shape=(x.shape[1:]),activation='relu',kernel_regularizer=regularizers.l2(weight_decay)))
model.add(layers.GlobalMaxPool1D())
#model.add(layers.Dense(num_hidden_unit,activation='relu'))
model.add(layers.Dense(num_hidden_unit,activation='relu',kernel_regularizer=regularizers.l2(weight_decay)))
model.add(layers.Dropout(0.5))
model.add(layers.Dense(1,activation='sigmoid'))
model.compile(optimizer=tf.keras.optimizers.RMSprop(),
loss='binary_crossentropy',
metrics=['acc'])
##4.模型训练
history = model.fit(x,y,
epochs=50,
batch_size=256,
validation_data=(self.x_validate,self.y_validate))
plt.plot(history.epoch,history.history.get('acc'),label='acc')
plt.plot(history.epoch,history.history.get('val_acc'),label='val_acc')
plt.legend()
self.model_retrained=model
model.save("D:/workspace of spyder/毕业设计/my project data/model_file/retrained_Binary_classification_"+self.p_file+".h5")
SE=sample_expansion("DNd41")
SE.expand_sample_slide(2)
SE.retrain_model()
print(SE.evaluation_loaded_model)
SE.model_retrained.evaluate(SE.x_test,SE.y_test)