def embeddingNoteVector(FilePath, DiDr, r, d):
'''
:param FilePath: 节点路径文件的路径
:param DiDr:游走完后的拼接矩阵
:return:
'''
embeddingVector_list = []
Pst_array = np.array(RFTL.readFileToList(FilePath, 1))
flag = False
if Pst_array.shape[0] != 0: #文件不为空
for i in range(Pst_array.shape[0]):
mid_list = []
if Pst_array[i][1] == r and Pst_array[i][3] == d:
mid_list.append([Pst_array[i][1],
Pst_array[i][3]]) # 起始药物ID和终止疾病ID
flag = True
if flag == True:
mid_list.append(DiDr[Pst_array[i][1]]) # 第一种药物
if Pst_array[i][0] == 0:
mid_list.append(DiDr[Pst_array[i][2]])
mid_list.append([0] * DiDr.shape[0])
mid_list.append([0] * DiDr.shape[0]) #最后元素补0
embeddingVector_list.append(mid_list)
if Pst_array[i][0] == 1: # r-r-d
mid_list.append(DiDr[Pst_array[i][2]]) #第二种药物
mid_list.append(DiDr[Pst_array[i][3] + 763]) #第一种疾病
mid_list.append([0] * DiDr.shape[0]) #最后元素补0
embeddingVector_list.append(mid_list)
if Pst_array[i][0] == 2: # r-d-d
mid_list.append(DiDr[Pst_array[i][2] + 763]) #第一种疾病
mid_list.append(DiDr[Pst_array[i][3] + 763]) #第二种疾病
mid_list.append([0] * DiDr.shape[0]) # 最后元素补0
embeddingVector_list.append(mid_list)
flag = False
return embeddingVector_list
else:
FPR_num += 1 / total_FPR
TPR.append(TPR_num)
FPR.append(FPR_num)
PRE.append(ipre_a / ipre_b)
TPR_Data.append(TPR)
FPR_Data.append(FPR)
PRE_Data.append(PRE)
return TPR_Data, FPR_Data, PRE_Data
x_FPR, y_TPR, y_PRE = [], [], []
# predict_list = np.load('..\data\Batch50E10LR0.001.pkl.npy')
predict_list = RFTL.readFileToList('..\data\prediction.txt', 0)
train_list = RFTL.readFileToList('..\data\第一份训练数据.txt', 1)
test_list = RFTL.readFileToList('..\data\第一份测试数据.txt', 1)
TPR, FPR, PRE = PlotRoc(np.array(predict_list), np.array(train_list),
np.array(test_list), [], [], [])
# TPR,FPR,PRE = PlotRoc(predict_list,np.array(train_list),np.array(test_list),[],[],[])
x_FPR.append(np.array(FPR).sum(axis=0) / np.array(FPR).shape[0])
y_TPR.append(np.array(TPR).sum(axis=0) / np.array(TPR).shape[0])
y_PRE.append(np.array(PRE).sum(axis=0) / np.array(PRE).shape[0])
print('ROC_AUC:', auc(x_FPR[0], y_TPR[0]))
print('PR_AUC:', auc(y_TPR[0], y_PRE[0]) + y_TPR[0][0] * y_PRE[0][0])
plot(x_FPR[0], y_TPR[0], "r")
plot(y_TPR[0], y_PRE[0], "b")
show()
x = self.conv2(x) #torch.Size([1, 32, 2, 348])
x = x.view(x.size(0), -1)
output = self.out1(x)
output = self.out2(output)
return output
cnn = CNN()
if torch.cuda.is_available():
cnn.cuda()
optimizer = torch.optim.Adam(cnn.parameters(), lr=0.001)
loss_func = nn.CrossEntropyLoss()
#--------------------------CNN模型结束-------------------------------------
EPOCH = 10
if __name__ == "__main__":
DrugSim_list = RFTL.readFileToList("..\data\drugSim.txt", 0)
DisSim_list = RFTL.readFileToList("..\data\disSim.txt", 0)
DiDr_list = RFTL.readFileToList('..\data\DiDrAMat.txt', 1)
DiDrSplit_list = RFTL.splitArray("..\data\DiDrAMat.txt") #所有1的坐标的list
DiDr_array, DiDr_testArray = RFTL.ChangeArray(np.array(DiDr_list),
DiDrSplit_list,
0) # 第一份1做test
np.savetxt('..\data\第一份训练数据_cnn.txt',
DiDr_array,
fmt=['%s'] * DiDr_array.shape[1],
newline='\n')
np.savetxt('..\data\第一份测试数据_cnn.txt',
DiDr_testArray,
fmt=['%s'] * DiDr_testArray.shape[1],
newline='\n')
#_Author_:Monkey
#!/usr/bin/env python
#-*- coding:utf-8 -*-
import numpy as np
import function as FUN
FileC_Path = '..\data\Batch50E10LR0.001.pkl.npy'
FileG_Path = '..\data\GRUM.txt'
cnn_list = np.load(FileC_Path)
gru_list = FUN.readFileToList(FileG_Path, 0)
predict = 0.6 * cnn_list + 0.4 * np.array(gru_list)
np.savetxt('..\data\prediction.txt',
predict,
fmt=['%s'] * predict.shape[1],
newline='\n')