-
Notifications
You must be signed in to change notification settings - Fork 0
/
dataset.py
228 lines (162 loc) · 11.5 KB
/
dataset.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
import random
import numpy as np
import pandas as pd
from sklearn.preprocessing import OrdinalEncoder
from sklearn.model_selection import train_test_split
import torch
from torch.utils.data import DataLoader, TensorDataset
import utils
import encoding_utils as eutils
import warnings
warnings.filterwarnings("ignore")
config = utils.get_config(print_dict = False)
seed = config["seed"]
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
np.random.seed(seed)
random.seed(seed)
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
def seed_worker(seed):
worker_seed = torch.manual_seed(seed)
np.random.seed(worker_seed)
random.seed(worker_seed)
G = torch.Generator()
G.manual_seed(seed)
config = utils.get_config(print_dict = False)
def data_prep(dataset, undataset):
dataset['bb1_skeleton'] = dataset.apply(lambda x:eutils.obtain_skeleton(x['bb1_smile'], bb_type = 3), axis=1)
dataset['bb2_skeleton'] = dataset.apply(lambda x:eutils.obtain_skeleton(x['bb2_smile'], bb_type = 2), axis=1)
dataset['collapsed'].fillna(value = 1, inplace = True)
dataset['bb2_replaced_smile'] = [eutils.double_to_single(smile) for smile in dataset['bb2_skeleton']]
undataset['bb2_replaced_smile'] = [eutils.double_to_single(smile) for smile in undataset['bb2_skeleton']]
dataset['bb2_smile_len'] = dataset.apply(lambda x: len(x['bb2_replaced_smile']), axis = 1)
undataset['bb2_smile_len'] = undataset.apply(lambda x: len(x['bb2_replaced_smile']), axis = 1)
smiles_new = dataset['bb2_replaced_smile']
un_smiles_new = undataset['bb2_replaced_smile']
max_len = config["max_len"]
smile_char_list = utils.load_vocab(config["vocab"])
data_prep.smile_to_index = dict((c,i) for i,c in enumerate(smile_char_list))
data_prep.index_to_smile = dict((i,c) for i,c in enumerate(smile_char_list))
vocab_length = eutils.smi_vocab_len(smile_char_list)
smiles_sos = np.array([eutils.add_sos(smile, mode = 'smile') for smile in smiles_new])
smiles_sos_eos = np.array([eutils.add_eos(smile, mode = 'smile') for smile in smiles_sos])
un_smiles_sos = np.array([eutils.add_sos(smile, mode = 'smile') for smile in un_smiles_new])
un_smiles_sos_eos = np.array([eutils.add_eos(smile, mode = 'smile') for smile in un_smiles_sos])
length_list = [len(smile) for smile in smiles_sos_eos]
smile2id_list = eutils.smiles_to_idx(smiles_sos_eos, max_len, data_prep.smile_to_index)
un_length_list = [len(smile) for smile in un_smiles_sos_eos]
un_smile2id_list = eutils.smiles_to_idx(un_smiles_sos_eos, max_len, data_prep.smile_to_index)
data_prep.ordinalenc = OrdinalEncoder()
ordinal_treat = pd.concat([dataset['bb1_skeleton'], dataset['reaction']], axis = 1)
data_prep.ordinalenc.fit(ordinal_treat)
features_list = data_prep.ordinalenc.transform(ordinal_treat)
un_ordinal_treat = pd.concat([undataset['bb1_skeleton'], undataset['reaction']], axis = 1)
un_features_list = data_prep.ordinalenc.transform(un_ordinal_treat)
collapse_label = dataset['collapsed']
undataset['collapsed'] = -1
uncollapse_label = undataset['collapsed']
dataset['indicator'] = 1
undataset['indicator'] = 0
# [:,:max_len], smile2id_list，
# [:, max_len], length_list，
# [:, max_len + 1:max_len + 3], features_list，
# [:, max_len + 3], collapse_label，
# [:, max_len + 4], indicator(shape persistency)
data_prep.supervised_input = np.hstack((np.array(smile2id_list),
np.array(length_list)[:, np.newaxis],
features_list,
np.array(collapse_label)[:, np.newaxis],
dataset['indicator'].to_numpy()[:, np.newaxis]))
data_prep.unsupervised_input = np.hstack((np.array(un_smile2id_list),
np.array(un_length_list)[:, np.newaxis],
un_features_list,
np.array(uncollapse_label)[:, np.newaxis],
undataset['indicator'].to_numpy()[:, np.newaxis]))
supervised_train, supervised_test = train_test_split(data_prep.supervised_input, test_size = 0.1, random_state = seed) #(32221, 57)， (1190304, 57)
unsupervised_train, unsupervised_test = train_test_split(data_prep.unsupervised_input, test_size = 0.002, random_state = seed) # (3581, 57)， (2386, 57)
# mix
data_prep.train_mixed_data = np.vstack((supervised_train, unsupervised_train))
data_prep.test_mixed_data = np.vstack((supervised_test, unsupervised_test))
# shuffle
np.random.shuffle(data_prep.train_mixed_data)
np.random.shuffle(data_prep.test_mixed_data)
train_data = torch.from_numpy(data_prep.train_mixed_data[:,:max_len]).type(torch.LongTensor)
test_data = torch.from_numpy(data_prep.test_mixed_data[:,:max_len]).type(torch.LongTensor)
train_len = torch.from_numpy(data_prep.train_mixed_data[:,max_len]).type(torch.LongTensor)
test_len = torch.from_numpy(data_prep.test_mixed_data[:,max_len]).type(torch.LongTensor)
train_features = torch.from_numpy(data_prep.train_mixed_data[:,max_len + 1:max_len + 3]).type(torch.LongTensor)
test_features = torch.from_numpy(data_prep.test_mixed_data[:,max_len + 1:max_len + 3]).type(torch.LongTensor)
train_target = torch.from_numpy(data_prep.train_mixed_data[:,max_len + 3]).type(torch.LongTensor)
test_target = torch.from_numpy(data_prep.test_mixed_data[:,max_len + 3]).type(torch.LongTensor)
train_indicator = torch.from_numpy(data_prep.train_mixed_data[:,max_len + 4]).type(torch.LongTensor)
test_indicator = torch.from_numpy(data_prep.test_mixed_data[:,max_len + 4]).type(torch.LongTensor)
all_train_data = TensorDataset(train_data, train_len, train_features, train_target, train_indicator)
all_test_data = TensorDataset(test_data, test_len, test_features, test_target, test_indicator)
# dataloader
train_dataloader = DataLoader(all_train_data, batch_size = 128, shuffle = True, worker_init_fn = seed_worker, generator = G)
test_dataloader = DataLoader(all_test_data, batch_size = len(all_test_data), shuffle = True, worker_init_fn = seed_worker, generator = G)
return train_dataloader, test_dataloader
import selfies as sf
def data_prep_selfie(dataset, undataset):
dataset['bb1_skeleton'] = dataset.apply(lambda x:eutils.obtain_skeleton(x['bb1_smile'], bb_type = 3), axis=1)
dataset['bb2_skeleton'] = dataset.apply(lambda x:eutils.obtain_skeleton(x['bb2_smile'], bb_type = 2), axis=1)
dataset['collapsed'].fillna(value = 1, inplace = True)
dataset['bb2_selfie_sk'] = eutils.to_selfies(dataset['bb2_skeleton'])
undataset['bb2_selfie_sk'] = eutils.to_selfies(undataset['bb2_skeleton'])
max_len = 57
selfie_char_list = utils.load_vocab("./vocab/selfie_vocab.npy")
data_prep_selfie.selfie_to_index = dict((c,i) for i,c in enumerate(selfie_char_list))
data_prep_selfie.index_to_selfie = dict((i,c) for i,c in enumerate(selfie_char_list))
vocab_length = eutils.selfie_vocab_len(selfie_char_list)
selfies_sos = np.array([eutils.add_sos(smile, mode = 'selfie') for smile in dataset['bb2_selfie_sk'] ])
selfies_sos_eos = np.array([eutils.add_eos(smile, mode = 'selfie') for smile in selfies_sos])
un_selfies_sos = np.array([eutils.add_sos(smile, mode = 'selfie') for smile in undataset['bb2_selfie_sk'] ])
un_selfies_sos_eos = np.array([eutils.add_eos(smile, mode = 'selfie') for smile in un_selfies_sos])
length_list = [len(smile) for smile in selfies_sos_eos]
selfie2id_list = eutils.selfies_to_idx(selfies_sos_eos, max_len, data_prep_selfie.selfie_to_index)
un_length_list = [len(smile) for smile in un_selfies_sos_eos]
un_selfie2id_list = eutils.selfies_to_idx(un_selfies_sos_eos, max_len, data_prep_selfie.selfie_to_index)
data_prep_selfie.ordinalenc = OrdinalEncoder()
ordinal_treat = pd.concat([dataset['bb1_skeleton'], dataset['reaction']], axis = 1)
data_prep_selfie.ordinalenc.fit(ordinal_treat)
features_list = data_prep_selfie.ordinalenc.transform(ordinal_treat)
un_ordinal_treat = pd.concat([undataset['bb1_skeleton'], undataset['reaction']], axis = 1)
un_features_list = data_prep_selfie.ordinalenc.transform(un_ordinal_treat)
collapse_label = dataset['collapsed']
undataset['collapsed'] = -1
uncollapse_label = undataset['collapsed']
dataset['indicator'] = 1
undataset['indicator'] = 0
data_prep_selfie.supervised_input = np.hstack((np.array(selfie2id_list),
np.array(length_list)[:, np.newaxis],
features_list,
np.array(collapse_label)[:, np.newaxis],
dataset['indicator'].to_numpy()[:, np.newaxis]))
data_prep_selfie.unsupervised_input = np.hstack((np.array(un_selfie2id_list),
np.array(un_length_list)[:, np.newaxis],
un_features_list,
np.array(uncollapse_label)[:, np.newaxis],
undataset['indicator'].to_numpy()[:, np.newaxis]))
supervised_train, supervised_test = train_test_split(data_prep_selfie.supervised_input, test_size = 0.1, random_state = seed)
unsupervised_train, unsupervised_test = train_test_split(data_prep_selfie.unsupervised_input, test_size = 0.002, random_state = seed)
data_prep_selfie.train_mixed_data = np.vstack((supervised_train, unsupervised_train))
data_prep_selfie.test_mixed_data = np.vstack((supervised_test, unsupervised_test))
np.random.shuffle(data_prep_selfie.train_mixed_data)
np.random.shuffle(data_prep_selfie.test_mixed_data)
train_data = torch.from_numpy(data_prep_selfie.train_mixed_data[:,:max_len]).type(torch.LongTensor)
test_data = torch.from_numpy(data_prep_selfie.test_mixed_data[:,:max_len]).type(torch.LongTensor)
train_len = torch.from_numpy(data_prep_selfie.train_mixed_data[:,max_len]).type(torch.LongTensor)
test_len = torch.from_numpy(data_prep_selfie.test_mixed_data[:,max_len]).type(torch.LongTensor)
train_features = torch.from_numpy(data_prep_selfie.train_mixed_data[:,max_len + 1:max_len + 3]).type(torch.LongTensor)
test_features = torch.from_numpy(data_prep_selfie.test_mixed_data[:,max_len + 1:max_len + 3]).type(torch.LongTensor)
train_target = torch.from_numpy(data_prep_selfie.train_mixed_data[:,max_len + 3]).type(torch.LongTensor)
test_target = torch.from_numpy(data_prep_selfie.test_mixed_data[:,max_len + 3]).type(torch.LongTensor)
train_indicator = torch.from_numpy(data_prep_selfie.train_mixed_data[:,max_len + 4]).type(torch.LongTensor)
test_indicator = torch.from_numpy(data_prep_selfie.test_mixed_data[:,max_len + 4]).type(torch.LongTensor)
all_train_data = TensorDataset(train_data, train_len, train_features, train_target, train_indicator)
all_test_data = TensorDataset(test_data, test_len, test_features, test_target, test_indicator)
train_dataloader = DataLoader(all_train_data, batch_size = 128, shuffle = True, worker_init_fn = seed_worker, generator = G)
test_dataloader = DataLoader(all_test_data, batch_size = len(all_test_data), shuffle = True, worker_init_fn = seed_worker, generator = G)
return train_dataloader, test_dataloader