add random seed parameter for splitter (47e3d28e) · Commits · 钟慕尧 / deepchem

deepchem/molnet/load_function/pdbbind_datasets.py

+7 −1

Original line number	Diff line number	Diff line
		@@ -144,6 +144,7 @@ def load_pdbbind(reload=True,
		load_binding_pocket=False,
		featurizer="grid",
		split="random",
		split_seed=None,
		save_dir=None):
		"""Load and featurize raw PDBBind dataset.

		@@ -312,6 +313,9 @@ def load_pdbbind(reload=True,
		print('Featurization complete.')
		# No transformations of data
		transformers = []

		# Split dataset
		print("\nSplit dataset...\n")
		if split == None:
		return pdbbind_tasks, (dataset, None, None), transformers

		@@ -322,7 +326,9 @@ def load_pdbbind(reload=True,
		'random': deepchem.splits.RandomSplitter(),
		}
		splitter = splitters[split]
		train, valid, test = splitter.train_valid_test_split(dataset)
		train, valid, test = splitter.train_valid_test_split(
		dataset, seed=split_seed)

		all_dataset = (train, valid, test)
		print("\nSaving dataset to \"%s\" ..." % save_folder)
		deepchem.utils.save.save_dataset_to_disk(save_folder, train, valid, test,