def test_compute_features_on_infinity_distance():
"""Test that WeaveModel correctly transforms WeaveMol objects into tensors with infinite max_pair_distance."""
featurizer = dc.feat.WeaveFeaturizer(max_pair_distance=None)
X = featurizer(["C", "CCC"])
batch_size = 20
model = WeaveModel(1,
batch_size=batch_size,
mode='classification',
fully_connected_layer_sizes=[2000, 1000],
batch_normalize=True,
batch_normalize_kwargs={
"fused": False,
"trainable": True,
"renorm": True
},
learning_rage=0.0005)
atom_feat, pair_feat, pair_split, atom_split, atom_to_pair = model.compute_features_on_batch(
X)
# There are 4 atoms each of which have 75 atom features
assert atom_feat.shape == (4, 75)
# There are 10 pairs with infinity distance and 14 pair features
assert pair_feat.shape == (10, 14)
# 4 atoms in total
assert atom_split.shape == (4, )
assert np.all(atom_split == np.array([0, 1, 1, 1]))
# 10 pairs in total
assert pair_split.shape == (10, )
assert np.all(pair_split == np.array([0, 1, 1, 1, 2, 2, 2, 3, 3, 3]))
# 10 pairs in total each with start/finish
assert atom_to_pair.shape == (10, 2)
assert np.all(
atom_to_pair == np.array([[0, 0], [1, 1], [1, 2], [1, 3], [2, 1],
[2, 2], [2, 3], [3, 1], [3, 2], [3, 3]]))
def test_weave_regression_model(self):
tasks, dataset, transformers, metric = self.get_dataset(
'regression', 'Weave')
model = WeaveModel(len(tasks), mode='regression')
model.fit(dataset, nb_epoch=80)
scores = model.evaluate(dataset, [metric], transformers)
assert all(s < 0.1 for s in scores['mean_absolute_error'])
def test_weave_fit_simple_distance_1():
featurizer = dc.feat.WeaveFeaturizer(max_pair_distance=1)
X = featurizer(["C", "CCC"])
y = np.array([0, 1.])
dataset = dc.data.NumpyDataset(X, y)
batch_size = 20
model = WeaveModel(1,
batch_size=batch_size,
mode='classification',
fully_connected_layer_sizes=[2000, 1000],
batch_normalize=True,
batch_normalize_kwargs={
"fused": False,
"trainable": True,
"renorm": True
},
learning_rage=0.0005)
model.fit(dataset, nb_epoch=200)
transformers = []
metric = dc.metrics.Metric(dc.metrics.roc_auc_score,
np.mean,
mode="classification")
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean-roc_auc_score'] >= 0.9
def test_weave_model(self):
tasks, dataset, transformers, metric = self.get_dataset(
'classification', 'Weave')
model = WeaveModel(len(tasks), mode='classification')
model.fit(dataset, nb_epoch=50)
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean-roc_auc_score'] >= 0.9
def test_weave_model():
tasks, dataset, transformers, metric = get_dataset(
'classification', 'Weave', data_points=10)
batch_size = 10
model = WeaveModel(
len(tasks),
batch_size=batch_size,
mode='classification',
dropouts=0,
learning_rate=0.0001)
model.fit(dataset, nb_epoch=250)
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean-roc_auc_score'] >= 0.9
def test_weave_regression_model():
import numpy as np
import tensorflow as tf
tf.random.set_seed(123)
np.random.seed(123)
tasks, dataset, transformers, metric = get_dataset('regression', 'Weave')
batch_size = 10
model = WeaveModel(len(tasks),
batch_size=batch_size,
mode='regression',
batch_normalize=False,
fully_connected_layer_sizes=[],
dropouts=0,
learning_rate=0.0005)
model.fit(dataset, nb_epoch=200)
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean_absolute_error'] < 0.1
def test_change_loss_function_weave(self):
tasks, dataset, transformers, metric = self.get_dataset(
'regression', 'Weave', num_tasks=1)
batch_size = 50
model = WeaveModel(
len(tasks), batch_size=batch_size, mode='regression', use_queue=False)
model.fit(dataset, nb_epoch=1)
model.save()
model2 = TensorGraph.load_from_dir(model.model_dir, restore=False)
dummy_label = model2.labels[-1]
dummy_ouput = model2.outputs[-1]
loss = ReduceSum(L2Loss(in_layers=[dummy_label, dummy_ouput]))
module = model2.create_submodel(loss=loss)
model2.restore()
model2.fit(dataset, nb_epoch=1, submodel=module)
def test_weave_model():
tasks, dataset, transformers, metric = get_dataset('classification',
'Weave')
batch_size = 20
model = WeaveModel(len(tasks),
batch_size=batch_size,
mode='classification',
fully_connected_layer_sizes=[2000, 1000],
batch_normalize=True,
batch_normalize_kwargs={
"fused": False,
"trainable": True,
"renorm": True
},
learning_rage=0.0005)
model.fit(dataset, nb_epoch=200)
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean-roc_auc_score'] >= 0.9
def test_compute_features_on_distance_1():
"""Test that WeaveModel correctly transforms WeaveMol objects into tensors with finite max_pair_distance."""
featurizer = dc.feat.WeaveFeaturizer(max_pair_distance=1)
X = featurizer(["C", "CCC"])
batch_size = 20
model = WeaveModel(
1,
batch_size=batch_size,
mode='classification',
fully_connected_layer_sizes=[2000, 1000],
batch_normalize=True,
batch_normalize_kwargs={
"fused": False,
"trainable": True,
"renorm": True
},
learning_rate=0.0005)
atom_feat, pair_feat, pair_split, atom_split, atom_to_pair = model.compute_features_on_batch(
X)
# There are 4 atoms each of which have 75 atom features
assert atom_feat.shape == (4, 75)
# There are 8 pairs with distance 1 and 14 pair features. (To see why 8,
# there's the self pair for "C". For "CCC" there are 7 pairs including self
# connections and accounting for symmetry.)
assert pair_feat.shape == (8, 14)
# 4 atoms in total
assert atom_split.shape == (4,)
assert np.all(atom_split == np.array([0, 1, 1, 1]))
# 10 pairs in total
assert pair_split.shape == (8,)
# The center atom is self connected and to both neighbors so it appears
# thrice. The canonical ranking used in MolecularFeaturizer means this
# central atom is ranked last in ordering.
assert np.all(pair_split == np.array([0, 1, 1, 2, 2, 3, 3, 3]))
# 10 pairs in total each with start/finish
assert atom_to_pair.shape == (8, 2)
assert np.all(atom_to_pair == np.array([[0, 0], [1, 1], [1, 3], [2, 2],
[2, 3], [3, 1], [3, 2], [3, 3]]))
def test_weave_regression_model(self):
tasks, dataset, transformers, metric = self.get_dataset(
'regression', 'Weave')
model = WeaveModel(len(tasks), mode='regression')
model.fit(dataset, nb_epoch=80)
scores = model.evaluate(dataset, [metric], transformers)
assert all(s < 0.1 for s in scores['mean_absolute_error'])
model.save()
model = TensorGraph.load_from_dir(model.model_dir)
scores2 = model.evaluate(dataset, [metric], transformers)
assert np.allclose(scores['mean_absolute_error'],
scores2['mean_absolute_error'])
def test_weave_model(self):
tasks, dataset, transformers, metric = self.get_dataset(
'classification', 'Weave')
model = WeaveModel(len(tasks), mode='classification')
model.fit(dataset, nb_epoch=50)
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean-roc_auc_score'] >= 0.9
model.save()
model = TensorGraph.load_from_dir(model.model_dir)
scores2 = model.evaluate(dataset, [metric], transformers)
assert np.allclose(scores['mean-roc_auc_score'],
scores2['mean-roc_auc_score'])
def test_weave_regression_model(self):
tasks, dataset, transformers, metric = self.get_dataset(
'regression', 'Weave')
model = WeaveModel(len(tasks), mode='regression')
model.fit(dataset, nb_epoch=80)
scores = model.evaluate(dataset, [metric], transformers)
assert all(s < 0.1 for s in scores['mean_absolute_error'])
model.save()
model = TensorGraph.load_from_dir(model.model_dir)
scores2 = model.evaluate(dataset, [metric], transformers)
assert np.allclose(scores['mean_absolute_error'],
scores2['mean_absolute_error'])
def test_weave_model(self):
tasks, dataset, transformers, metric = self.get_dataset(
'classification', 'Weave')
model = WeaveModel(len(tasks), mode='classification')
model.fit(dataset, nb_epoch=50)
scores = model.evaluate(dataset, [metric], transformers)
assert scores['mean-roc_auc_score'] >= 0.9
model.save()
model = TensorGraph.load_from_dir(model.model_dir)
scores2 = model.evaluate(dataset, [metric], transformers)
assert np.allclose(scores['mean-roc_auc_score'],
scores2['mean-roc_auc_score'])
def test_change_loss_function_weave(self):
tasks, dataset, transformers, metric = self.get_dataset(
'regression', 'Weave', num_tasks=1)
batch_size = 50
model = WeaveModel(
len(tasks), batch_size=batch_size, mode='regression', use_queue=False)
model.fit(dataset, nb_epoch=1)
model.save()
model2 = TensorGraph.load_from_dir(model.model_dir, restore=False)
dummy_label = model2.labels[-1]
dummy_ouput = model2.outputs[-1]
loss = ReduceSum(L2Loss(in_layers=[dummy_label, dummy_ouput]))
module = model2.create_submodel(loss=loss)
model2.restore()
model2.fit(dataset, nb_epoch=1, submodel=module)
def generate_weave_model():
batch_size = 64
model = WeaveModel(1, batch_size=batch_size, learning_rate=1e-3, use_queue=False, mode='regression')
return model
metric = dc.metrics.Metric(dc.metrics.pearson_r2_score, np.mean)
# Do setup required for tf/keras models
n_tasks = 1 #Only solubility to predict
n_atom_feat = 75
n_pair_feat = 14
n_hidden = 10
batch_size = 64
n_graph_feat = 10
nb_epoch = 10
model = WeaveModel(
n_tasks=n_tasks,
n_atom_feat=n_atom_feat,
n_pair_feat=n_pair_feat,
n_hidden=n_hidden,
n_graph_feat=n_graph_feat,
mode="regression",
batch_size=batch_size,
model_dir=
"/home/rod/Dropbox/Quimica/Analysis/ANalisis/Borradores/WeaveModel/"
) #To prevent overfitting
# Fit trained model
print("Fitting model")
model.fit(train_dataset, nb_epoch=nb_epoch)
model.save()
print("Evaluating model")
train_scores = model.evaluate(train_dataset, [metric], transformers)
valid_scores = model.evaluate(valid_dataset, [metric], transformers)
print("Train scores")
