-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathNormalCV_EXP.py
193 lines (171 loc) · 7.01 KB
/
NormalCV_EXP.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
import pandas as pd
import numpy as np
import os
import random
import sys
import math
from tensorflow import keras
from tensorflow.keras.models import Sequential
from sklearn.model_selection import train_test_split
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint
#from tensorflow.keras.layers.convolutional import Conv2D
#from tensorflow.keras.layers.convolutional import MaxPooling2D
from tensorflow.keras.layers import Dense, Dropout, Input, Activation,Flatten
from sklearn.metrics import mean_squared_error
#from keras import backend as K
import scipy.stats as stats
from sklearn.preprocessing import StandardScaler, MinMaxScaler
from tensorflow.keras import models, layers
from sklearn.metrics import r2_score,mean_absolute_error
import tensorflow as tf
from tensorflow import keras
from tensorflow.python.keras import backend as K
from tensorflow.python.ops import array_ops
from tensorflow.python.ops import math_ops
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import KFold
#Please customize this path
def pearson(pred,y):
pear = stats.pearsonr(y, pred)
pear_value = pear[0]
pear_p_val = pear[1]
print("Pearson correlation is {} and related p_value is {}".format(pear_value, pear_p_val))
return pear_value
def spearman(pred,y):
spear = stats.spearmanr(y, pred)
spear_value = spear[0]
spear_p_val = spear[1]
print("Spearman correlation is {} and related p_value is {}".format(spear_value, spear_p_val))
return spear_value
def DNN(inputShape,n1=1024,n2=1024,n3=500,lr=0.0001):
model = models.Sequential()
model.add(layers.Dense(n1,kernel_initializer="he_normal", input_shape=[inputShape]))
model.add(layers.Dropout(0.5))# % of features dropped)
#model.add(layers.Dense(1024, activation='relu',kernel_initializer="he_normal"))
#model.add(layers.Dropout(0.3))# % of features dropped)
model.add(layers.Dense(n2, activation='relu',kernel_initializer="he_normal"))
model.add(layers.Dropout(0.3))# % of features dropped)
model.add(layers.Dense(n3, activation='tanh',kernel_initializer="he_normal"))
# output layer
model.add(layers.Dense(1))
model.compile( optimizer=keras.optimizers.Adam(learning_rate=float(lr),beta_1=0.9, beta_2=0.999, amsgrad=False), loss='mean_squared_error',metrics=['mse', 'mae'])
return model
#Please customize this path
data=pd.read_csv('/homes/rzgar/Narjes/Inputs/DrugComb.csv')
FeatureName=['A1','A2','A3','A4','A5','B1','B2','B3','B4','B5','C1','C2','C3','C4','C5','D1','D2','D3','D4','D5','E1','E2','E3','E4','E5']
def concatRefine(x_train,y_train):
x_new=np.zeros((x_train.shape[0]*2,x_train.shape[1]))
y_new=[]
i=-1
for x,y in zip(x_train,y_train):
i=i+1
x_new[i,:]=np.concatenate((x[128:256],x[0:128],x[256:356]),axis=0)
y_new.append(y)
i=i+1
x_new[i,:]=x
y_new.append(y)
return x_new,y_new
def creatCombUNiq(data):
UniqList=[]
for d1, d2 in zip(data.Drug1,data.Drug2):
if((d2+'//'+d1) not in UniqList):
UniqList.append(d1+'//'+d2)
UniqList=np.unique(np.array(UniqList))
return UniqList
def concatPandas(data,out):
NewTrain=pd.DataFrame()
for d1, d2 in zip(out[:,0],out[:,1]):
TrainName=data[(((data.Drug1==d1) & (data.Drug2==d2 ))|((data.Drug1==d2) & (data.Drug2==d1 )))]
NewTrain = pd.concat([NewTrain,TrainName])
return NewTrain
def toNUMPY(d):
data=[]
for i in d:
data.append(np.array(i))
return np.array(data)
def runapp(i,n1,n2,n3,batch,lr,part):
f = open("/nfs/research/petsalaki/users/rzgar/Outputs/EXP_Normal_"+str(FeatureName[i])+".txt", "a")
rmse_mean=0
sp_mean=0
pcc_mean=0
#Please customize this paths
X=np.load('/nfs/research/petsalaki/users/rzgar/Inputs/FeaturesR1.npy', mmap_mode='r')[:,:,i]
y=np.load('/nfs/research/petsalaki/users/rzgar/Inputs/ScoresR1.npy', mmap_mode='r')[:,2]
kf = KFold(n_splits=5, random_state=94, shuffle=True)
kf.get_n_splits(X)
prenew=np.array([])
realnew=np.array([])
i=0
outFinal=pd.DataFrame()
for train_index, test_index in kf.split(X):
outPut=pd.DataFrame()
i=i+1
X_train, X_test = X[train_index], X[test_index]
y_train, y_test = y[train_index], y[test_index]
X_train, X_val, y_train, y_val = train_test_split( X_train, y_train, test_size=0.20,random_state=42)
X_train,y_train=concatRefine(X_train,y_train)
outPut['fold']=np.zeros([len(test_index),])+i
outPut['Index']=test_index
CNN_model=DNN(356,n1,n2,n3,lr)
#Please customize these paths
cb_check = ModelCheckpoint(('/nfs/research/petsalaki/users/rzgar/Outputs/DNN_EXPNormal_'+str(part)+'_'+str(FeatureName[i])+'_Loewe'), verbose=1, monitor='val_loss',save_best_only=True, mode='auto')
y_train=np.array(y_train)
CNN_model.fit(x=X_train,y=y_train,batch_size=batch,epochs = 100,shuffle=True,validation_data = (X_val,y_val),callbacks=[EarlyStopping(monitor='val_loss', mode='auto', patience = 10),cb_check] )
CNN_model = tf.keras.models.load_model('/nfs/research/petsalaki/users/rzgar/Outputs/DNN_EXPNormal_'+str(part)+'_'+str(FeatureName[i])+'_Loewe')
pre_test=CNN_model.predict(X_test)
pcc=pearson(pre_test[:,0],y_test)
sp=spearman(pre_test,y_test)
rmse=math.sqrt( mean_squared_error(y_test ,pre_test))
rmse_mean=rmse+rmse_mean
sp_mean=sp+sp_mean
pcc_mean=pcc_mean+pcc
prenew=np.concatenate((pre_test,prenew), axis=None)
realnew=np.concatenate((y_test,realnew), axis=None)
outPut['Real']=y_test
outPut['Pre']=pre_test
outFinal = pd.concat([outFinal,outPut])
# np.save('/nfs/research/petsalaki/users/rzgar/Normal_EXP_Pre_'+str(FeatureName[i])+'.npy', prenew)
#
# np.save('/nfs/research/petsalaki/users/rzgar/Normal_EXP_Real_'+str(FeatureName[i])+'.npy', realnew)
outFinal.to_csv('/nfs/research/petsalaki/users/rzgar/Result_NormalCV_EXP_'+str(FeatureName[i])+'.csv')
rmse_mean=rmse_mean/5
sp_mean=sp_mean/5
pcc_mean=pcc_mean/5
f.write(str(FeatureName[i])+ ' sp: '+str(sp)+' rmse:'+str(rmse)+' pc: '+str(pcc))
f.write('\n')
f.write('Metrics after concat:')
f.write('\n')
pcc=pearson(prenew[:,0],realnew)
sp=spearman(prenew[:,0],realnew)
rmse=math.sqrt( mean_squared_error(prenew[:,0],realnew))
f.write(str(FeatureName[i])+ ' sp: '+str(sp)+' rmse:'+str(rmse)+' pc: '+str(pcc))
f.write('\n')
f.flush()
f.close()
def main():
i=0
n1=1024
n2=1024
n3=500
lr=0.0001
batch=128
part=1
# get the options from user
for arg in sys.argv[1:]:
(key,val) = arg.rstrip().split('=')
if key == 'DrugF':
i=val
if key == 'N1':
n1=val
if key == 'N2':
n2=val
if key == 'N3':
n3=val
if key == 'Batch':
batch=val
if key == 'LR':
lr=val
if key == 'Part':
part=val
runapp(int(i),int(n1),int(n2),int(n3),int(batch),float(lr),part)
main()