/
uv_datasets.py
197 lines (144 loc) · 6.63 KB
/
uv_datasets.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
"""
UV Dataset loader
"""
import os
import logging
import time
import numpy as np
import deepchem
from deepchem.molnet.load_function.kaggle_features import merck_descriptors
from deepchem.molnet.load_function.uv_tasks import UV_tasks
logger = logging.getLogger(__name__)
TRAIN_URL = "https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/UV_training_disguised_combined_full.csv.gz"
VALID_URL = "https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/UV_test1_disguised_combined_full.csv.gz"
TEST_URL = "https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/UV_test2_disguised_combined_full.csv.gz"
TRAIN_FILENAME = "UV_training_disguised_combined_full.csv.gz"
VALID_FILENAME = "UV_test1_disguised_combined_full.csv.gz"
TEST_FILENAME = "UV_test2_disguised_combined_full.csv.gz"
def remove_missing_entries(dataset):
"""Remove missing entries.
Some of the datasets have missing entries that sneak in as zero'd out
feature vectors. Get rid of them.
"""
for i, (X, y, w, ids) in enumerate(dataset.itershards()):
available_rows = X.any(axis=1)
logger.info("Shard %d has %d missing entries." %
(i, np.count_nonzero(~available_rows)))
X = X[available_rows]
y = y[available_rows]
w = w[available_rows]
ids = ids[available_rows]
dataset.set_shard(i, X, y, w, ids)
def get_transformers(train_dataset):
"Gets transformations applied on the dataset"
transformers = list()
return transformers
def gen_uv(UV_tasks, data_dir, train_dir, valid_dir, test_dir, shard_size=2000):
"""Loading the UV dataset; does not do train/test split"""
time1 = time.time()
train_files = os.path.join(data_dir, TRAIN_FILENAME)
valid_files = os.path.join(data_dir, VALID_FILENAME)
test_files = os.path.join(data_dir, TEST_FILENAME)
# Download files if they don't exist
if not os.path.exists(train_files):
logger.info("Downloading training file...")
deepchem.utils.data_utils.download_url(url=TRAIN_URL, dest_dir=data_dir)
logger.info("Training file download complete.")
logger.info("Downloading validation file...")
deepchem.utils.data_utils.download_url(url=VALID_URL, dest_dir=data_dir)
logger.info("Validation file download complete.")
logger.info("Downloading test file...")
deepchem.utils.data_utils.download_url(url=TEST_URL, dest_dir=data_dir)
logger.info("Test file download complete")
# Featurizing datasets
logger.info("About to featurize UV dataset.")
featurizer = deepchem.feat.UserDefinedFeaturizer(merck_descriptors)
loader = deepchem.data.UserCSVLoader(
tasks=UV_tasks, id_field="Molecule", featurizer=featurizer)
logger.info("Featurizing train datasets...")
train_dataset = loader.featurize(
input_files=train_files, shard_size=shard_size)
logger.info("Featurizing validation datasets...")
valid_dataset = loader.featurize(
input_files=valid_files, shard_size=shard_size)
logger.info("Featurizing test datasets....")
test_dataset = loader.featurize(input_files=test_files, shard_size=shard_size)
# Missing entry removal
logger.info("Removing missing entries from dataset.")
remove_missing_entries(train_dataset)
remove_missing_entries(valid_dataset)
remove_missing_entries(test_dataset)
# Shuffle the training data
logger.info("Shuffling the training dataset")
train_dataset.sparse_shuffle()
# Apply transformations
logger.info("Starting transformations")
transformers = get_transformers(train_dataset)
for transformer in transformers:
logger.info("Performing transformations with {}".format(
transformer.__class__.__name__))
logger.info("Transforming the training dataset...")
train_dataset = transformer.transform(train_dataset)
logger.info("Transforming the validation dataset...")
valid_dataset = transformer.transform(valid_dataset)
logger.info("Transforming the test dataset...")
test_dataset = transformer.transform(test_dataset)
logger.info("Transformations complete.")
logger.info("Moving datasets to corresponding directories")
train_dataset.move(train_dir)
logger.info("Train dataset moved.")
valid_dataset.move(valid_dir)
logger.info("Validation dataset moved.")
test_dataset.move(test_dir)
logger.info("Test dataset moved.")
time2 = time.time()
##### TIMING ###########
logger.info("TIMING: UV fitting took %0.3f s" % (time2 - time1))
return train_dataset, valid_dataset, test_dataset
def load_uv(shard_size=2000, featurizer=None, split=None, reload=True):
"""Load UV dataset; does not do train/test split
The UV dataset is an in-house dataset from Merck that was first introduced in the following paper:
Ramsundar, Bharath, et al. "Is multitask deep learning practical for pharma?." Journal of chemical information and modeling 57.8 (2017): 2068-2076.
The UV dataset tests 10,000 of Merck's internal compounds on
190 absorption wavelengths between 210 and 400 nm. Unlike
most of the other datasets featured in MoleculeNet, the UV
collection does not have structures for the compounds tested
since they were proprietary Merck compounds. However, the
collection does feature pre-computed descriptors for these
compounds.
Note that the original train/valid/test split from the source
data was preserved here, so this function doesn't allow for
alternate modes of splitting. Similarly, since the source data
came pre-featurized, it is not possible to apply alternative
featurizations.
Parameters
----------
shard_size: int, optional
Size of the DiskDataset shards to write on disk
featurizer: optional
Ignored since featurization pre-computed
split: optional
Ignored since split pre-computed
reload: bool, optional
Whether to automatically re-load from disk
"""
data_dir = deepchem.utils.data_utils.get_data_dir()
data_dir = os.path.join(data_dir, "UV")
if not os.path.exists(data_dir):
os.mkdir(data_dir)
train_dir = os.path.join(data_dir, "train_dir")
valid_dir = os.path.join(data_dir, "valid_dir")
test_dir = os.path.join(data_dir, "test_dir")
if (os.path.exists(train_dir) and os.path.exists(valid_dir) and
os.path.exists(test_dir)):
logger.info("Reloading existing datasets")
train_dataset = deepchem.data.DiskDataset(train_dir)
valid_dataset = deepchem.data.DiskDataset(valid_dir)
test_dataset = deepchem.data.DiskDataset(test_dir)
else:
logger.info("Featurizing datasets")
train_dataset, valid_dataset, test_dataset = \
gen_uv(UV_tasks=UV_tasks, data_dir=data_dir, train_dir=train_dir,
valid_dir=valid_dir, test_dir=test_dir, shard_size=shard_size)
transformers = get_transformers(train_dataset)
return UV_tasks, (train_dataset, valid_dataset, test_dataset), transformers