def setUp(self):
self.temp_file = TempSMILESFile(tempfile_kwargs={'prefix': 'dataset'})
self.fh = self.temp_file.open()
self.item_list = self.temp_file.smiles_strings.split('\n')
self.dataset = SMILESDataset(self.fh.name)
def setUp(self):
self.temp_file = TempSMILESFile()
self.fh = self.temp_file.open()
# See `test_data.py` for data set test cases.
self.dataset = SMILESDataset(self.fh.name)
self.vocab = SMILESVocabulary(self.dataset, need_corpus=True)
def setUp(self):
temp_file = TempSMILESFile(
tempfile_kwargs={'prefix': 'softmax_sampler'})
self.fh = temp_file.open()
dataset = SMILESDataset(self.fh.name)
self.vocabulary = SMILESVocabulary(dataset, need_corpus=True)
self.model = SMILESRNN(len(self.vocabulary))
self.predictor = SoftmaxSearch(self.model, self.vocabulary)
class SMILESDatasetTestCase(unittest.TestCase):
def setUp(self):
self.temp_file = TempSMILESFile(tempfile_kwargs={'prefix': 'dataset'})
self.fh = self.temp_file.open()
self.item_list = self.temp_file.smiles_strings.split('\n')
self.dataset = SMILESDataset(self.fh.name)
def test_1_read(self):
self.assertTrue(
all(
s.startswith(Token.BOS) and s.endswith(Token.EOS)
for s in self.dataset))
self.assertListEqual(
self.item_list,
[Token.crop(s) for s in self.dataset],
)
self.assertEqual(
len(self.item_list),
len(self.dataset),
)
def test_2_map_(self):
mapped_data_iter = self.dataset.map_(Token.crop)
self.assertIsInstance(mapped_data_iter, Iterator)
self.assertListEqual(self.item_list, list(mapped_data_iter))
def test_3_filter_(self):
filtered_data_iter = self.dataset.filter_(lambda s: len(s) < 10)
self.assertIsInstance(filtered_data_iter, Iterator)
self.assertListEqual(['{N#N}', '{CN=C=O}'], list(filtered_data_iter))
def tearDown(self):
self.fh.close()
def setUp(self):
self.temp_file = TempSMILESFile(
tempfile_kwargs={'prefix': 'dataloader'})
self.fh = self.temp_file.open()
dataset = SMILESDataset(self.fh.name)
vocabulary = SMILESVocabulary(dataset=dataset, need_corpus=True)
self.dataloader = SMILESBatchColumnSampler(
corpus=vocabulary.corpus,
batch_size=2,
n_steps=4,
shuffle=True,
)
def setUp(self):
self.smiles_strings = (
'CC(=O)NCCC1=CNc2c1cc(OC)cc2\n'
'CCc1c[n+]2ccc3c4ccccc4[nH]c3c2cc1\n'
'O1C=C[C@H]([C@H]1O2)c3c2cc(OC)c4c3OC(=O)C5=C4CCC(=O)5'
)
self.fh = tempfile.NamedTemporaryFile(mode='w+', encoding='ascii')
self.fh.write(self.smiles_strings)
self.fh.seek(os.SEEK_SET)
# See `test_data.py` for data set test cases.
self.dataset = SMILESDataset(self.fh.name)
self.vocab = Vocabulary(self.dataset, need_corpus=True)
def setUp(self):
temp_file = TempSMILESFile(tempfile_kwargs={'prefix': 'model'})
self.fh = temp_file.open()
dataset = SMILESDataset(self.fh.name)
self.vocabulary = SMILESVocabulary(dataset, need_corpus=True)
self.batch_sampler = SMILESBatchColumnSampler(
corpus=self.vocabulary.corpus,
batch_size=3,
n_steps=8,
)
self.n_rnn_layers = 1 # Used in output/state shape testing.
self.n_rnn_units = 32 # Used in output/state shape testing.
self.model = SMILESRNN(len(self.vocabulary),
use_one_hot=False,
embedding_dim=4,
embedding_dropout=0.25,
embedding_dropout_axes=0,
embedding_init=mx.init.Uniform(),
embedding_prefix='embedding_',
rnn='lstm',
rnn_n_layers=self.n_rnn_layers,
rnn_n_units=self.n_rnn_units,
rnn_i2h_init='xavier_normal',
rnn_h2h_init='orthogonal_normal',
rnn_reinit_state=True,
rnn_detach_state=False,
rnn_state_init=mx.nd.random.uniform,
rnn_dropout=0.0,
rnn_prefix='encoder_',
dense_n_layers=2,
dense_n_units=32,
dense_activation='relu',
dense_dropout=0.5,
dense_init=mx.init.Xavier(),
dense_prefix='decoder_',
dtype='float32',
prefix='model_')
def setUp(self):
self.smiles_string = 'CCc1c[n+]2ccc3c4ccccc4[nH]c3c2cc1'
with TempSMILESFile(smiles_strings=self.smiles_string) as temp_fh:
dataset = SMILESDataset(temp_fh.file_handler.name)
self.vocabulary = SMILESVocabulary(dataset, need_corpus=True)