Loading README.md +63 −52 Original line number Diff line number Diff line Loading @@ -244,7 +244,7 @@ Scaffold splitting * Regression |Dataset |Model |Splitting |Train score/R2|Valid score/R2| |-----------|--------------------|------------|--------------|--------------| |----------------|--------------------|------------|--------------|--------------| |delaney |MT-NN regression |Index |0.773 |0.574 | | |graphconv regression|Index |0.991 |0.825 | | |MT-NN regression |Random |0.769 |0.591 | Loading @@ -257,6 +257,9 @@ Scaffold splitting | |graphconv regression|Random |0.117 |0.076 | | |MT-NN regression |Scaffold |0.180 |0.052 | | |graphconv regression|Scaffold |0.131 |0.046 | |pdbbind(core) |MT-NN regression |Random |0.973 |0.494 | |pdbbind(refined)|MT-NN regression |Random |0.987 |0.503 | |pdbbind(full) |MT-NN regression |Random |0.983 |0.528 | |kaggle |MT-NN regression |User-defined|0.748 |0.452 | * General features Loading @@ -264,7 +267,7 @@ Scaffold splitting Number of tasks and examples in the datasets |Dataset |N(tasks) |N(samples) | |-----------|-----------|-----------| |----------------|-----------|-----------| |tox21 |12 |8014 | |muv |17 |93127 | |pcba |128 |439863 | Loading @@ -273,11 +276,16 @@ Number of tasks and examples in the datasets |delaney |1 |1128 | |kaggle |15 |173065 | |nci |60 |19127 | |pdbbind(core) |1 |195 | |pdbbind(refined)|1 |3706 | |pdbbind(full) |1 |11908 | Time needed for benchmark test(~20h in total) |Dataset |Model |Time(loading)/s |Time(running)/s| |-----------|--------------------|----------------|---------------| |----------------|--------------------|----------------|---------------| |tox21 |logistic regression |30 |60 | | |Multitask network |30 |60 | | |robust MT-NN |30 |90 | Loading @@ -302,6 +310,9 @@ Time needed for benchmark test(~20h in total) | |graphconv regression|10 |40 | |nci |MT-NN regression |400 |1200 | | |graphconv regression|400 |2500 | |pdbbind(core) |MT-NN regression |0(featurized) |30 | |pdbbind(refined)|MT-NN regression |0(featurized) |40 | |pdbbind(full) |MT-NN regression |0(featurized) |60 | |kaggle |MT-NN regression |2200 |3200 | Loading deepchem/splits/splitters.py +4 −2 Original line number Diff line number Diff line Loading @@ -269,6 +269,7 @@ class MolecularWeightSplitter(Splitter): """ np.testing.assert_almost_equal(frac_train + frac_valid + frac_test, 1.) if not seed is None: np.random.seed(seed) mws = [] Loading Loading @@ -299,6 +300,7 @@ class RandomSplitter(Splitter): Splits internal compounds randomly into train/validation/test. """ np.testing.assert_almost_equal(frac_train + frac_valid + frac_test, 1.) if not seed is None: np.random.seed(seed) num_datapoints = len(dataset) train_cutoff = int(frac_train * num_datapoints) Loading examples/benchmark.py +21 −9 Original line number Diff line number Diff line Loading @@ -17,7 +17,7 @@ on datasets: muv, pcba, tox21, sider, toxcast Giving regression performances of: MultitaskDNN(tf_regression), Graph convolution regression(graphconvreg) on datasets: delaney, nci, kaggle on datasets: delaney, nci, kaggle, pdbbind time estimation listed in README file Loading Loading @@ -49,6 +49,7 @@ from sider.sider_datasets import load_sider from kaggle.kaggle_datasets import load_kaggle from delaney.delaney_datasets import load_delaney from nci.nci_datasets import load_nci from pdbbind.pdbbind_datasets import load_pdbbind_grid def benchmark_loading_datasets(hyper_parameters, dataset='tox21', model='tf', split=None, Loading @@ -74,7 +75,7 @@ def benchmark_loading_datasets(hyper_parameters, if dataset in ['muv', 'pcba', 'tox21', 'sider', 'toxcast']: mode = 'classification' elif dataset in ['kaggle', 'delaney', 'nci']: elif dataset in ['kaggle', 'delaney', 'nci','pdbbind']: mode = 'regression' else: raise ValueError('Dataset not supported') Loading @@ -91,7 +92,19 @@ def benchmark_loading_datasets(hyper_parameters, if dataset in ['kaggle']: featurizer = None #kaggle dataset use its own features if split in ['random', 'scaffold']: return else: split = None #kaggle dataset is already splitted if not model in ['tf_regression']: return if dataset in ['pdbbind']: featurizer = 'grid' #pdbbind use grid featurizer if split in ['scaffold', 'index']: return #skip the scaffold and index splitting of pdbbind if not model in ['tf_regression']: return if not split in [None, 'index','random','scaffold']: raise ValueError('Splitter function not supported') Loading @@ -99,7 +112,8 @@ def benchmark_loading_datasets(hyper_parameters, loading_functions = {'tox21': load_tox21, 'muv': load_muv, 'pcba': load_pcba, 'nci': load_nci, 'sider': load_sider, 'toxcast': load_toxcast, 'kaggle': load_kaggle, 'delaney': load_delaney} 'kaggle': load_kaggle, 'delaney': load_delaney, 'pdbbind': load_pdbbind_grid} print('-------------------------------------') print('Benchmark %s on dataset: %s' % (model, dataset)) Loading @@ -117,7 +131,7 @@ def benchmark_loading_datasets(hyper_parameters, train_dataset, valid_dataset, test_dataset = all_dataset time_finish_loading = time.time() #time_finish_loading-time_start is the time(s) used for dataset loading if dataset in ['kaggle']: if dataset in ['kaggle','pdbbind']: n_features = train_dataset.get_data_shape()[0] #kaggle dataset has customized features Loading Loading @@ -497,7 +511,7 @@ if __name__ == '__main__': 'tf_regression, graphconvreg') parser.add_argument('-d', action='append', dest='dataset_args', default=[], help='Choice of dataset: tox21, sider, muv, toxcast, pcba, ' + 'kaggle, delaney, nci') 'kaggle, delaney, nci, pdbbind') args = parser.parse_args() #Datasets and models used in the benchmark test splitters = args.splitter_args Loading @@ -511,7 +525,7 @@ if __name__ == '__main__': 'tf_regression', 'graphconvreg'] if len(datasets) == 0: datasets = ['tox21', 'sider', 'muv', 'toxcast', 'pcba', 'delaney', 'nci', 'kaggle'] 'delaney', 'nci', 'kaggle', 'pdbbind'] #input hyperparameters #tf: dropouts, learning rate, layer_sizes, weight initial stddev,penalty, Loading Loading @@ -562,8 +576,6 @@ if __name__ == '__main__': benchmark_loading_datasets( hps, dataset=dataset, model=model, split=split, out_path='.') else: if dataset in ['kaggle']: datasets.remove('kaggle') #kaggle only needs to be run once for model in models: if model in ['tf_regression', 'graphconvreg']: benchmark_loading_datasets( Loading examples/gdb7/gdb7_datasets.py 0 → 100644 +42 −0 Original line number Diff line number Diff line """ gdb7 dataset loader. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals import os import numpy as np import shutil import deepchem as dc def load_gdb7(featurizer=None, split='index'): """Load gdb7 datasets.""" # Featurize gdb7 dataset print("About to featurize gdb7 dataset.") current_dir = os.path.dirname(os.path.realpath(__file__)) dataset_file = os.path.join( current_dir, "./gdb7.sdf") gdb7_tasks = ["u0_atom"] if featurizer is None: featurizer = dc.feat.CoulombMatrixEig(23) else: raise ValueError('Only support Coulomb Matrix featurizer') loader = dc.data.SDFLoader(tasks=gdb7_tasks, smiles_field="smiles", mol_field="mol", featurizer=featurizer) dataset = loader.featurize(dataset_file) # Initialize transformers transformers = [ dc.trans.NormalizationTransformer(transform_X=True, dataset=dataset), dc.trans.NormalizationTransformer(transform_y=True, dataset=dataset)] print("About to transform data") for transformer in transformers: dataset = transformer.transform(dataset) splitters = {'index': dc.splits.IndexSplitter(), 'random': dc.splits.RandomSplitter()} splitter = splitters[split] train, valid, test = splitter.train_valid_test_split(dataset) return gdb7_tasks, (train, valid, test), transformers examples/gdb7/gdb7_tf2.py 0 → 100644 +37 −0 Original line number Diff line number Diff line """ Script that trains Tensorflow singletask models on GDB7 dataset. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals import os import deepchem as dc import numpy as np from gdb7_datasets import load_gdb7 np.random.seed(123) gdb7_tasks, datasets, transformers = load_gdb7() train_dataset, valid_dataset, test_dataset = datasets regression_metric = dc.metrics.Metric(dc.metrics.mean_absolute_error, mode="regression") model = dc.models.TensorflowMultiTaskRegressor( n_tasks=len(gdb7_tasks), n_features=23, learning_rate=.001, momentum=.8, batch_size=512, weight_init_stddevs=[1/np.sqrt(2000),1/np.sqrt(800),1/np.sqrt(800),1/np.sqrt(1000)], bias_init_consts=[0.,0.,0.,0.], layer_sizes=[2000,800,800,1000], dropouts=[0.1,0.1,0.1,0.1]) # Fit trained model model.fit(train_dataset) model.save() train_scores = model.evaluate(train_dataset, [regression_metric], transformers) print("Train scores [kcal/mol]") print(train_scores) valid_scores = model.evaluate(valid_dataset, [regression_metric], transformers) print("Validation scores [kcal/mol]") print(valid_scores) Loading
README.md +63 −52 Original line number Diff line number Diff line Loading @@ -244,7 +244,7 @@ Scaffold splitting * Regression |Dataset |Model |Splitting |Train score/R2|Valid score/R2| |-----------|--------------------|------------|--------------|--------------| |----------------|--------------------|------------|--------------|--------------| |delaney |MT-NN regression |Index |0.773 |0.574 | | |graphconv regression|Index |0.991 |0.825 | | |MT-NN regression |Random |0.769 |0.591 | Loading @@ -257,6 +257,9 @@ Scaffold splitting | |graphconv regression|Random |0.117 |0.076 | | |MT-NN regression |Scaffold |0.180 |0.052 | | |graphconv regression|Scaffold |0.131 |0.046 | |pdbbind(core) |MT-NN regression |Random |0.973 |0.494 | |pdbbind(refined)|MT-NN regression |Random |0.987 |0.503 | |pdbbind(full) |MT-NN regression |Random |0.983 |0.528 | |kaggle |MT-NN regression |User-defined|0.748 |0.452 | * General features Loading @@ -264,7 +267,7 @@ Scaffold splitting Number of tasks and examples in the datasets |Dataset |N(tasks) |N(samples) | |-----------|-----------|-----------| |----------------|-----------|-----------| |tox21 |12 |8014 | |muv |17 |93127 | |pcba |128 |439863 | Loading @@ -273,11 +276,16 @@ Number of tasks and examples in the datasets |delaney |1 |1128 | |kaggle |15 |173065 | |nci |60 |19127 | |pdbbind(core) |1 |195 | |pdbbind(refined)|1 |3706 | |pdbbind(full) |1 |11908 | Time needed for benchmark test(~20h in total) |Dataset |Model |Time(loading)/s |Time(running)/s| |-----------|--------------------|----------------|---------------| |----------------|--------------------|----------------|---------------| |tox21 |logistic regression |30 |60 | | |Multitask network |30 |60 | | |robust MT-NN |30 |90 | Loading @@ -302,6 +310,9 @@ Time needed for benchmark test(~20h in total) | |graphconv regression|10 |40 | |nci |MT-NN regression |400 |1200 | | |graphconv regression|400 |2500 | |pdbbind(core) |MT-NN regression |0(featurized) |30 | |pdbbind(refined)|MT-NN regression |0(featurized) |40 | |pdbbind(full) |MT-NN regression |0(featurized) |60 | |kaggle |MT-NN regression |2200 |3200 | Loading
deepchem/splits/splitters.py +4 −2 Original line number Diff line number Diff line Loading @@ -269,6 +269,7 @@ class MolecularWeightSplitter(Splitter): """ np.testing.assert_almost_equal(frac_train + frac_valid + frac_test, 1.) if not seed is None: np.random.seed(seed) mws = [] Loading Loading @@ -299,6 +300,7 @@ class RandomSplitter(Splitter): Splits internal compounds randomly into train/validation/test. """ np.testing.assert_almost_equal(frac_train + frac_valid + frac_test, 1.) if not seed is None: np.random.seed(seed) num_datapoints = len(dataset) train_cutoff = int(frac_train * num_datapoints) Loading
examples/benchmark.py +21 −9 Original line number Diff line number Diff line Loading @@ -17,7 +17,7 @@ on datasets: muv, pcba, tox21, sider, toxcast Giving regression performances of: MultitaskDNN(tf_regression), Graph convolution regression(graphconvreg) on datasets: delaney, nci, kaggle on datasets: delaney, nci, kaggle, pdbbind time estimation listed in README file Loading Loading @@ -49,6 +49,7 @@ from sider.sider_datasets import load_sider from kaggle.kaggle_datasets import load_kaggle from delaney.delaney_datasets import load_delaney from nci.nci_datasets import load_nci from pdbbind.pdbbind_datasets import load_pdbbind_grid def benchmark_loading_datasets(hyper_parameters, dataset='tox21', model='tf', split=None, Loading @@ -74,7 +75,7 @@ def benchmark_loading_datasets(hyper_parameters, if dataset in ['muv', 'pcba', 'tox21', 'sider', 'toxcast']: mode = 'classification' elif dataset in ['kaggle', 'delaney', 'nci']: elif dataset in ['kaggle', 'delaney', 'nci','pdbbind']: mode = 'regression' else: raise ValueError('Dataset not supported') Loading @@ -91,7 +92,19 @@ def benchmark_loading_datasets(hyper_parameters, if dataset in ['kaggle']: featurizer = None #kaggle dataset use its own features if split in ['random', 'scaffold']: return else: split = None #kaggle dataset is already splitted if not model in ['tf_regression']: return if dataset in ['pdbbind']: featurizer = 'grid' #pdbbind use grid featurizer if split in ['scaffold', 'index']: return #skip the scaffold and index splitting of pdbbind if not model in ['tf_regression']: return if not split in [None, 'index','random','scaffold']: raise ValueError('Splitter function not supported') Loading @@ -99,7 +112,8 @@ def benchmark_loading_datasets(hyper_parameters, loading_functions = {'tox21': load_tox21, 'muv': load_muv, 'pcba': load_pcba, 'nci': load_nci, 'sider': load_sider, 'toxcast': load_toxcast, 'kaggle': load_kaggle, 'delaney': load_delaney} 'kaggle': load_kaggle, 'delaney': load_delaney, 'pdbbind': load_pdbbind_grid} print('-------------------------------------') print('Benchmark %s on dataset: %s' % (model, dataset)) Loading @@ -117,7 +131,7 @@ def benchmark_loading_datasets(hyper_parameters, train_dataset, valid_dataset, test_dataset = all_dataset time_finish_loading = time.time() #time_finish_loading-time_start is the time(s) used for dataset loading if dataset in ['kaggle']: if dataset in ['kaggle','pdbbind']: n_features = train_dataset.get_data_shape()[0] #kaggle dataset has customized features Loading Loading @@ -497,7 +511,7 @@ if __name__ == '__main__': 'tf_regression, graphconvreg') parser.add_argument('-d', action='append', dest='dataset_args', default=[], help='Choice of dataset: tox21, sider, muv, toxcast, pcba, ' + 'kaggle, delaney, nci') 'kaggle, delaney, nci, pdbbind') args = parser.parse_args() #Datasets and models used in the benchmark test splitters = args.splitter_args Loading @@ -511,7 +525,7 @@ if __name__ == '__main__': 'tf_regression', 'graphconvreg'] if len(datasets) == 0: datasets = ['tox21', 'sider', 'muv', 'toxcast', 'pcba', 'delaney', 'nci', 'kaggle'] 'delaney', 'nci', 'kaggle', 'pdbbind'] #input hyperparameters #tf: dropouts, learning rate, layer_sizes, weight initial stddev,penalty, Loading Loading @@ -562,8 +576,6 @@ if __name__ == '__main__': benchmark_loading_datasets( hps, dataset=dataset, model=model, split=split, out_path='.') else: if dataset in ['kaggle']: datasets.remove('kaggle') #kaggle only needs to be run once for model in models: if model in ['tf_regression', 'graphconvreg']: benchmark_loading_datasets( Loading
examples/gdb7/gdb7_datasets.py 0 → 100644 +42 −0 Original line number Diff line number Diff line """ gdb7 dataset loader. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals import os import numpy as np import shutil import deepchem as dc def load_gdb7(featurizer=None, split='index'): """Load gdb7 datasets.""" # Featurize gdb7 dataset print("About to featurize gdb7 dataset.") current_dir = os.path.dirname(os.path.realpath(__file__)) dataset_file = os.path.join( current_dir, "./gdb7.sdf") gdb7_tasks = ["u0_atom"] if featurizer is None: featurizer = dc.feat.CoulombMatrixEig(23) else: raise ValueError('Only support Coulomb Matrix featurizer') loader = dc.data.SDFLoader(tasks=gdb7_tasks, smiles_field="smiles", mol_field="mol", featurizer=featurizer) dataset = loader.featurize(dataset_file) # Initialize transformers transformers = [ dc.trans.NormalizationTransformer(transform_X=True, dataset=dataset), dc.trans.NormalizationTransformer(transform_y=True, dataset=dataset)] print("About to transform data") for transformer in transformers: dataset = transformer.transform(dataset) splitters = {'index': dc.splits.IndexSplitter(), 'random': dc.splits.RandomSplitter()} splitter = splitters[split] train, valid, test = splitter.train_valid_test_split(dataset) return gdb7_tasks, (train, valid, test), transformers
examples/gdb7/gdb7_tf2.py 0 → 100644 +37 −0 Original line number Diff line number Diff line """ Script that trains Tensorflow singletask models on GDB7 dataset. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals import os import deepchem as dc import numpy as np from gdb7_datasets import load_gdb7 np.random.seed(123) gdb7_tasks, datasets, transformers = load_gdb7() train_dataset, valid_dataset, test_dataset = datasets regression_metric = dc.metrics.Metric(dc.metrics.mean_absolute_error, mode="regression") model = dc.models.TensorflowMultiTaskRegressor( n_tasks=len(gdb7_tasks), n_features=23, learning_rate=.001, momentum=.8, batch_size=512, weight_init_stddevs=[1/np.sqrt(2000),1/np.sqrt(800),1/np.sqrt(800),1/np.sqrt(1000)], bias_init_consts=[0.,0.,0.,0.], layer_sizes=[2000,800,800,1000], dropouts=[0.1,0.1,0.1,0.1]) # Fit trained model model.fit(train_dataset) model.save() train_scores = model.evaluate(train_dataset, [regression_metric], transformers) print("Train scores [kcal/mol]") print(train_scores) valid_scores = model.evaluate(valid_dataset, [regression_metric], transformers) print("Validation scores [kcal/mol]") print(valid_scores)
