-
Notifications
You must be signed in to change notification settings - Fork 0
/
PlncRNA-HDeep.py
291 lines (241 loc) · 9.99 KB
/
PlncRNA-HDeep.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
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
# PlncRNA-HDeep
import numpy as np
import re
import math
import nltk
import collections
from keras.utils import np_utils
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Convolution2D, MaxPooling2D, Flatten, normalization, Bidirectional
from keras import optimizers
from keras.layers.embeddings import Embedding
from keras.layers.recurrent import LSTM
from keras.preprocessing import sequence
from sklearn.model_selection import train_test_split
from sklearn import metrics
from sklearn import preprocessing
np.random.seed(1337) # random seed
# Evaluation criteria
def comparison(testlabel, resultslabel):
TP = 0
FP = 0
TN = 0
FN = 0
for row1 in range(len(resultslabel)):
if resultslabel[row1] < 0.5:
resultslabel[row1] = 0
else:
resultslabel[row1] = 1
for row2 in range(len(testlabel)):
if testlabel[row2] == 1 and testlabel[row2] == resultslabel[row2]:
TP = TP + 1
if testlabel[row2] == 0 and testlabel[row2] != resultslabel[row2]:
FP = FP + 1
if testlabel[row2] == 0 and testlabel[row2] == resultslabel[row2]:
TN = TN + 1
if testlabel[row2] == 1 and testlabel[row2] != resultslabel[row2]:
FN = FN + 1
# TPR:sensitivity, recall, hit rate or true positive rate
if TP + FN != 0:
TPR = TP / (TP + FN)
else:
TPR = 999999
# TNR:specificity, selectivity or true negative rate
if TN + FP != 0:
TNR = TN / (TN + FP)
else:
TNR = 999999
# PPV:precision or positive predictive value
if TP + FP != 0:
PPV = TP / (TP + FP)
else:
PPV = 999999
# NPV:negative predictive value
if TN + FN != 0:
NPV = TN / (TN + FN)
else:
NPV = 999999
# FNR:miss rate or false negative rate
if FN + TP != 0:
FNR = FN / (FN + TP)
else:
FNR = 999999
# FPR:fall-out or false positive rate
if FP + TN != 0:
FPR = FP / (FP + TN)
else:
FPR = 999999
# FDR:false discovery rate
if FP + TP != 0:
FDR = FP / (FP + TP)
else:
FDR = 999999
# FOR:false omission rate
if FN + TN != 0:
FOR = FN / (FN + TN)
else:
FOR = 999999
# ACC:accuracy
if TP + TN + FP + FN != 0:
ACC = (TP + TN) / (TP + TN + FP + FN)
else:
ACC = 999999
# F1 score:is the harmonic mean of precision and sensitivity
if TP + FP + FN != 0:
F1 = (2 * TP) / (2 * TP + FP + FN)
else:
F1 = 999999
# MCC:Matthews correlation coefficient
if (TP + FP) * (TP + FN) * (TN + FP) * (TN + FN) != 0:
MCC = (TP * TN + FP * FN) / math.sqrt((TP + FP) * (TP + FN) * (TN + FP) * (TN + FN))
else:
MCC = 999999
# BM:Informedness or Bookmaker Informedness
if TPR != 999999 and TNR != 999999:
BM = TPR + TNR - 1
else:
BM = 999999
# MK:Markedness
if PPV != 999999 and NPV != 999999:
MK = PPV + NPV - 1
else:
MK = 999999
return TP, FP, TN, FN, TPR, TNR, PPV, NPV, FNR, FPR, FDR, FOR, ACC, F1, MCC, BM, MK
filepath = 'Datasets\\TotalDataset.fasta'
## lncRNA-LSTM using p-nucleotide encoding ##############################################################################################################################
maxlen = 0 # max sequence length
word_freqs = collections.Counter() # word frequency
num_recs = 0 # number of samples
k = 3 # p-nucleotide encoding
with open(filepath, 'r+', encoding='gb18030', errors='ignore') as f:
for line in f:
name, sentence, label = line.strip().split(",")
sentence = sentence.lower()
words=[]
count=0
while(count<len(sentence)):
if(len(sentence[count:count+k])==k):
words.append(sentence[count:count + k]) # ['tact','actg']
count += k
if len(words) > maxlen:
maxlen = len(words)
for word in words:
word_freqs[word] += 1
num_recs += 1
vocab_size = min(64, len(word_freqs)) + 2
word2index = {x[0]: i+2 for i, x in enumerate(word_freqs.most_common(64))}
word2index["PAD"] = 0
word2index["UNK"] = 1
index2word = {v: k for k, v in word2index.items()}
X = np.empty(num_recs, dtype=list)
y = np.zeros(num_recs)
i = 0
with open(filepath, 'r+', encoding='gb18030', errors='ignore') as f:
for line in f:
name, sentence, label = line.strip().split(",")
sentence = sentence.lower()
words=[]
count=0
while(count<len(sentence)):
if(len(sentence[count:count+k])==k):
words.append(sentence[count:count + k]) # ['tact','actg']
count += k
seqs = []
for word in words:
if word in word2index:
seqs.append(word2index[word])
else:
seqs.append(word2index["UNK"])
X[i] = seqs
y[i] = int(label)
i += 1
X = sequence.pad_sequences(X, maxlen=3100, padding='post')
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1337)
# Building the lncRNA-LSTM model
model = Sequential()
model.add(Embedding(vocab_size, 128, input_length=3100))
model.add(Bidirectional(LSTM(64, dropout=0.4)))
model.add(Dense(1, activation="sigmoid"))
model.compile(loss="binary_crossentropy", optimizer="adam", metrics=["accuracy"])
model.summary()
# Training
model.fit(X_train, y_train, batch_size=128, epochs=3)
# Prediction
loss, accuracy = model.evaluate(X_test, y_test, batch_size=128)
resultslabelLSTM = model.predict(X_test)
## lncRNA-LSTM using p-nucleotide encoding ###############################################################################################################################
## CNN using one-hot encoding ############################################################################################################################################
list = open(filepath, 'r').readlines()
threshold = 0 # max sequence length
for linelength in list:
name, sequence, label = linelength.split(',')
if len(sequence) > threshold:
threshold = len(sequence)
# one-hot encoding
def onehot(list, threshold):
onehotsequence = []
onehotlabel = []
ATCG = 'ATCG'
char_to_int = dict((c, j) for j, c in enumerate(ATCG))
for line in list:
name, sequence, label = line.split(',')
rstr = r"[BDEFHIJKLMNOPQRSVWXYZ]"
sequence = re.sub(rstr, '', sequence)
integer_encoded = [char_to_int[char] for char in sequence]
hot_encoded = []
for value in integer_encoded:
letter = [0 for _ in range(len(ATCG))]
letter[value] = 1
hot_encoded.append(letter)
# zero-padding
if len(hot_encoded) < threshold:
zero = threshold - len(hot_encoded)
letter = [0 for _ in range(len(ATCG))]
for i in range(zero):
hot_encoded.append(letter)
hot_encoded_array = np.array(hot_encoded).reshape(-1, 4)
onehotsequence.append(hot_encoded_array)
onehotlabel.append(label.strip('\n'))
X = np.array(onehotsequence).reshape(-1, threshold, 4, 1)
X = X.astype('float32')
y = np.array(onehotlabel).astype('int').reshape(-1, 1)
y = np_utils.to_categorical(y, num_classes=2)
return X, y
X2, y2 = onehot(list, threshold)
traindata, testdata, trainlabel, testlabel = train_test_split(X2, y2, test_size=0.2, random_state=1337)
# Building the CNN model
model = Sequential()
model.add(Convolution2D(batch_input_shape=(None, threshold, 4, 1), filters=32, kernel_size=4, strides=1, padding='same', data_format='channels_last'))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=4, strides=4, padding='same', data_format='channels_last'))
model.add(Convolution2D(64, 4, strides=1, padding='same', data_format='channels_first'))
model.add(Activation('relu'))
model.add(MaxPooling2D(4, 4, 'same', data_format='channels_last'))
model.add(Flatten())
model.add(Dense(1024))
model.add(Activation('relu'))
model.add(Dropout(0.4))
model.add(Dense(2))
model.add(Activation('softmax'))
sgd = optimizers.SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(optimizer=sgd, loss='categorical_crossentropy', metrics=['accuracy'])
# Training
print('Training --------------')
model.fit(traindata, trainlabel, epochs=10, batch_size=128, verbose=1)
# Prediction
loss, accuracy = model.evaluate(testdata, testlabel, batch_size=128)
resultslabelCNN = model.predict(testdata)
## CNN using one-hot encoding ############################################################################################################################################
## Hybrid deep learning using three strategies ###########################################################################################################################
for rowfuz in range(resultslabelCNN.shape[0]):
# if abs(2 * resultslabelLSTM[rowfuz] - 1) < abs(2 * resultslabelCNN[rowfuz][1] - 1): # greedy hybrid strategy
# if abs(2 * resultslabelCNN[rowfuz][1] - 1) >= 0.5: # predominance of CNN hybrid strategy
if abs(2 * resultslabelLSTM[rowfuz] - 1) < 0.5: # predominance of lncRNA-LSTM hybrid strategy
resultslabelLSTM[rowfuz] = resultslabelCNN[rowfuz][1]
## Hybrid deep learning using three strategies ###########################################################################################################################
# Evaluation
TP, FP, TN, FN, TPR, TNR, PPV, NPV, FNR, FPR, FDR, FOR, ACC, F1, MCC, BM, MK = comparison(y_test, resultslabelLSTM)
print('PlncRNAPred')
print('TP:', TP, 'FP:', FP, 'TN:', TN, 'FN:', FN)
print('TPR:', TPR, 'TNR:', TNR, 'PPV:', PPV, 'NPV:', NPV, 'FNR:', FNR, 'FPR:', FPR, 'FDR:', FDR, 'FOR:', FOR)
print('ACC:', ACC, 'F1:', F1, 'MCC:', MCC, 'BM:', BM, 'MK:', MK)