-
Notifications
You must be signed in to change notification settings - Fork 5
/
data.py
300 lines (271 loc) · 10.2 KB
/
data.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
292
293
294
295
296
297
298
299
300
import numpy as np
import os.path
import subprocess
import casphandle
import utils
TRAIN_PATH = 'data/cullpdb+profile_6133_filtered.npy.gz'
TEST_PATH = 'data/cb513+profile_split1.npy.gz'
##### TRAIN DATA #####
def get_train(seq_len=None):
if not os.path.isfile(TRAIN_PATH):
print("Train path is not downloaded ...")
subprocess.call("./download_train.sh", shell=True)
else:
print("Train path is downloaded ...")
print("Loading train data ...")
X_in = utils.load_gz(TRAIN_PATH)
X = np.reshape(X_in,(5534,700,57))
del X_in
X = X[:,:,:]
labels = X[:,:,22:30]
mask = X[:,:,30] * -1 + 1
a = np.arange(0,21)
b = np.arange(35,56)
c = np.hstack((a,b))
X = X[:,:,c]
# getting meta
num_seqs = np.size(X,0)
seqlen = np.size(X,1)
d = np.size(X,2)
num_classes = 8
#### REMAKING LABELS ####
X = X.astype("float32")
mask = mask.astype("float32")
# Dummy -> concat
vals = np.arange(0,8)
labels_new = np.zeros((num_seqs,seqlen))
for i in range(np.size(labels,axis=0)):
labels_new[i,:] = np.dot(labels[i,:,:], vals)
labels_new = labels_new.astype('int32')
labels = labels_new
print("Loading splits ...")
##### SPLITS #####
# getting splits (cannot run before splits are made)
#split = np.load("data/split.pkl")
seq_names = np.arange(0,num_seqs)
#np.random.shuffle(seq_names)
X_train = X[seq_names[0:5278]]
X_valid = X[seq_names[5278:5534]]
labels_train = labels[seq_names[0:5278]]
labels_valid = labels[seq_names[5278:5534]]
mask_train = mask[seq_names[0:5278]]
mask_valid = mask[seq_names[5278:5534]]
num_seq_train = np.size(X_train,0)
num_seq_valid = np.size(X_valid,0)
if seq_len is not None:
X_train = X_train[:, :seq_len]
X_valid = X_valid[:, :seq_len]
labels_train = labels_train[:, :seq_len]
labels_valid = labels_valid[:, :seq_len]
mask_train = mask_train[:, :seq_len]
mask_valid = mask_valid[:, :seq_len]
len_train = np.sum(mask_train, axis=1)
len_valid = np.sum(mask_valid, axis=1)
return X_train, X_valid, labels_train, labels_valid, mask_train, \
mask_valid, len_train, len_valid, num_seq_train
#del split
##### TEST DATA #####
def get_test(seq_len=None):
if not os.path.isfile(TEST_PATH):
subprocess.call("./download_test.sh", shell=True)
print("Loading test data ...")
X_test_in = utils.load_gz(TEST_PATH)
X_test = np.reshape(X_test_in,(514,700,57))
del X_test_in
X_test = X_test[:,:,:].astype("float32")
labels_test = X_test[:,:,22:30].astype('int32')
mask_test = X_test[:,:,30].astype("float32") * -1 + 1
a = np.arange(0,21)
b = np.arange(35,56)
c = np.hstack((a,b))
X_test = X_test[:,:,c]
# getting meta
seqlen = np.size(X_test,1)
d = np.size(X_test,2)
num_classes = 8
num_seq_test = np.size(X_test,0)
del a, b, c
## DUMMY -> CONCAT ##
vals = np.arange(0,8)
labels_new = np.zeros((num_seq_test,seqlen))
for i in range(np.size(labels_test,axis=0)):
labels_new[i,:] = np.dot(labels_test[i,:,:], vals)
labels_new = labels_new.astype('int32')
labels_test = labels_new
### ADDING BATCH PADDING ###
X_add = np.zeros((126,seqlen,d))
label_add = np.zeros((126,seqlen))
mask_add = np.zeros((126,seqlen))
#
X_test = np.concatenate((X_test,X_add), axis=0).astype("float32")
labels_test = np.concatenate((labels_test, label_add), axis=0).astype('int32')
mask_test = np.concatenate((mask_test, mask_add), axis=0).astype("float32")
if seq_len is not None:
X_test = X_test[:, :seq_len]
labels_test = labels_test[:, :seq_len]
mask_test = mask_test[:, :seq_len]
len_test = np.sum(mask_test, axis=1)
len_test[-126:] = np.ones((126,), dtype='int32')
return X_test, mask_test, labels_test, num_seq_test, len_test
def get_casp(seq_len=None):
X_casp, t_casp, mask_casp = casphandle.get_data()
# getting meta
seqlen = np.size(X_casp,1)
d = np.size(X_casp,2)
num_classes = 8
### ADDING BATCH PADDING ###
num_add = 256 - X_casp.shape[0]
X_add = np.zeros((num_add,seqlen,d))
t_add = np.zeros((num_add,seqlen))
mask_add = np.zeros((num_add,seqlen))
#
X_casp = np.concatenate((X_casp, X_add), axis=0).astype("float32")
t_casp = np.concatenate((t_casp, t_add), axis=0).astype('int32')
mask_casp = np.concatenate((mask_casp, mask_add), axis=0).astype("float32")
if seq_len is not None:
X_casp = X_casp[:, :seq_len]
t_casp = t_casp[:, :seq_len]
mask_casp = mask_casp[:, :seq_len]
len_casp = np.sum(mask_casp, axis=1)
len_casp[-num_add:] = np.ones((num_add,), dtype='int32')
return X_casp, mask_casp, t_casp, len_casp
def load_data():
X_train, X_valid, t_train, t_valid, mask_train, \
mask_valid, len_train, len_valid, num_seq_train = get_train()
X_test, mask_test, t_test, num_seq_test, len_test = get_test()
X_casp, mask_casp, t_casp, len_casp = get_casp()
dict_out = dict()
dict_out['X_train'] = X_train
dict_out['X_valid'] = X_valid
dict_out['X_test'] = X_test
dict_out['X_casp'] = X_casp
dict_out['t_train'] = t_train
dict_out['t_valid'] = t_valid
dict_out['t_test'] = t_test
dict_out['t_casp'] = t_casp
dict_out['mask_train'] = mask_train
dict_out['mask_valid'] = mask_valid
dict_out['mask_test'] = mask_test
dict_out['mask_casp'] = mask_casp
dict_out['length_train'] = len_train
dict_out['length_valid'] = len_valid
dict_out['length_test'] = len_test
dict_out['length_casp'] = len_casp
return dict_out
class gen_data():
def __init__(self, num_iterations=1000001, batch_size=64, data_fn=load_data):
print("initializing data generator!")
self._num_iterations = num_iterations
self._batch_size = batch_size
self._data_dict = load_data()
self._seq_len = 700
print(self._data_dict.keys())
if 'X_train' in self._data_dict.keys():
if 't_train' in self._data_dict.keys():
print("Training is found!")
self._idcs_train = list(range(self._data_dict['X_train'].shape[0]))
self._num_features = self._data_dict['X_train'].shape[-1]
if 'X_valid' in self._data_dict.keys():
if 't_valid' in self._data_dict.keys():
print("Valid is found!")
self._idcs_valid = list(range(self._data_dict['X_valid'].shape[0]))
if 'X_test' in self._data_dict.keys():
if 't_test' in self._data_dict.keys():
print("Test is found!")
self._idcs_test = list(range(self._data_dict['X_test'].shape[0]))
if 'X_casp' in self._data_dict.keys():
if 't_casp' in self._data_dict.keys():
print("CASP is found!")
self._idcs_casp = list(range(self._data_dict['X_casp'].shape[0]))
def _shuffle_train(self):
np.random.shuffle(self._idcs_train)
def _batch_init(self):
batch_holder = dict()
batch_holder["X"] = np.zeros((self._batch_size, self._seq_len, self._num_features), dtype="float32")
batch_holder["t"] = np.zeros((self._batch_size, self._seq_len), dtype="int32")
batch_holder["mask"] = np.zeros((self._batch_size, self._seq_len), dtype="float32")
batch_holder["length"] = np.zeros((self._batch_size,), dtype="int32")
return batch_holder
def _chop_batch(self, batch, i=None):
X, t, mask = utils.chop_sequences(batch['X'], batch['t'], batch['mask'], batch['length'])
if i is None:
batch['X'] = X
batch['t'] = t
batch['mask'] = mask
else:
batch['X'] = X[:i]
batch['t'] = t[:i]
batch['mask'] = mask[:i]
return batch
def gen_valid(self):
batch = self._batch_init()
i = 0
for idx in self._idcs_valid:
batch['X'][i] = self._data_dict['X_valid'][idx]
batch['t'][i] = self._data_dict['t_valid'][idx]
batch['mask'][i] = self._data_dict['mask_valid'][idx]
batch['length'][i] = self._data_dict['length_valid'][idx]
i += 1
if i >= self._batch_size:
yield self._chop_batch(batch, i), i
batch = self._batch_init()
i = 0
if i != 0:
yield self._chop_batch(batch, i), i
def gen_test(self):
batch = self._batch_init()
i = 0
for idx in self._idcs_test:
batch['X'][i] = self._data_dict['X_test'][idx]
batch['t'][i] = self._data_dict['t_test'][idx]
batch['mask'][i] = self._data_dict['mask_test'][idx]
batch['length'][i] = self._data_dict['length_test'][idx]
i += 1
if i >= self._batch_size:
yield self._chop_batch(batch, i), i
batch = self._batch_init()
i = 0
if i != 0:
print(i)
print(self._chop_batch(batch, i)['X'].shape)
yield self._chop_batch(batch, i), i
def gen_casp(self):
batch = self._batch_init()
i = 0
for idx in self._idcs_casp:
batch['X'][i] = self._data_dict['X_casp'][idx]
batch['t'][i] = self._data_dict['t_casp'][idx]
batch['mask'][i] = self._data_dict['mask_casp'][idx]
batch['length'][i] = self._data_dict['length_casp'][idx]
i += 1
if i >= self._batch_size:
yield self._chop_batch(batch, i), i
batch = self._batch_init()
i = 0
if i != 0:
print(i)
print(self._chop_batch(batch, i)['X'].shape)
yield self._chop_batch(batch, i), i
def gen_train(self):
batch = self._batch_init()
iteration = 0
i = 0
while True:
# shuffling all batches
self._shuffle_train()
for idx in self._idcs_train:
batch['X'][i] = self._data_dict['X_train'][idx]
batch['t'][i] = self._data_dict['t_train'][idx]
batch['mask'][i] = self._data_dict['mask_train'][idx]
batch['length'][i] = self._data_dict['length_train'][idx]
i += 1
if i >= self._batch_size:
yield self._chop_batch(batch)
batch = self._batch_init()
i = 0
iteration += 1
if iteration >= self._num_iterations:
break
else:
continue
break