Local improvements. (3895d0fd) · Commits · 钟慕尧 / deepchem

deep_chem/models/deep.py

+6 −5

Original line number	Diff line number	Diff line
		@@ -146,10 +146,6 @@ def fit_multitask_mlp(paths, task_types, task_transforms, desc_transforms,
		r2s = compute_r2_scores(results, local_task_types)
		if r2s:
		print "Mean R^2: %f" % np.mean(np.array(r2s.values()))
		#rms = compute_rms_scores(results, local_task_types)
		#if rms:
		# print "Mean RMS: %f" % np.mean(np.array(rms.values()))
		#return (aucs, r2s, rms)

		def fit_singletask_mlp(paths, task_types, task_transforms,
		desc_transforms, splittype="random",
		@@ -177,7 +173,9 @@ def fit_singletask_mlp(paths, task_types, task_transforms,
		desc_weight=desc_weight)
		ret_vals = {}
		aucs, r2s, rms = {}, {}, {}
		for index, target in enumerate(singletasks):
		sorted_targets = sorted(singletasks.keys())
		for index, target in enumerate(sorted_targets):
		print "Training model %d" % index
		(train, X_train, y_train, W_train, test, X_test, y_test, W_test) = (
		singletasks[target])
		model = train_multitask_model(X_train, y_train, W_train,
		@@ -198,10 +196,13 @@ def fit_singletask_mlp(paths, task_types, task_transforms,
		r2s.update(target_r2s)
		rms.update(target_rms)
		if aucs:
		print aucs
		print "Mean AUC: %f" % np.mean(np.array(aucs.values()))
		if r2s:
		print r2s
		print "Mean R^2: %f" % np.mean(np.array(r2s.values()))
		if rms:
		print rms
		print "Mean RMS: %f" % np.mean(np.array(rms.values()))

		def train_multitask_model(X, y, W, task_types, desc_transforms, add_descriptors=False,

deep_chem/models/standard.py

+6 −1

Original line number	Diff line number	Diff line
		@@ -2,6 +2,7 @@
		Code for processing datasets using scikit-learn.
		"""
		import numpy as np
		from deep_chem.utils.analysis import results_to_csv
		from deep_chem.utils.load import load_and_transform_dataset
		from deep_chem.utils.preprocess import multitask_to_singletask
		from deep_chem.utils.preprocess import train_test_random_split
		@@ -51,7 +52,8 @@ def fit_singletask_models(paths, modeltype, task_types, task_transforms,
		add_descriptors=add_descriptors)
		singletask = multitask_to_singletask(dataset)
		aucs, r2s, rms = {}, {}, {}
		for target in singletask:
		for index, target in enumerate(sorted(singletask.keys())):
		print "Building model %d" % index
		data = singletask[target]
		if splittype == "random":
		train, test = train_test_random_split(data, seed=seed)
		@@ -94,10 +96,13 @@ def fit_singletask_models(paths, modeltype, task_types, task_transforms,
		r2s.update(target_r2s)
		rms.update(target_rms)
		if aucs:
		print results_to_csv(aucs)
		print "Mean AUC: %f" % np.mean(np.array(aucs.values()))
		if r2s:
		print results_to_csv(r2s)
		print "Mean R^2: %f" % np.mean(np.array(r2s.values()))
		if rms:
		print results_to_csv(rms)
		print "Mean RMS: %f" % np.mean(np.array(rms.values()))

deep_chem/scripts/launch_models.py

+1 −1

Original line number	Diff line number	Diff line
		@@ -79,7 +79,7 @@ def main():
		nb_epoch=args.n_epochs, decay=args.decay, validation_split=args.validation_split)
		else:
		fit_singletask_models(paths.values(), args.model, task_types,
		task_transforms, splittype="scaffold")
		task_transforms, splittype=args.splittype)

		if __name__ == "__main__":
		main()

deep_chem/utils/analysis.py

+6 −0

Original line number	Diff line number	Diff line
		@@ -7,6 +7,12 @@ __license__ = "LGPL"

		import numpy as np

		def results_to_csv(results_dict):
		"""Pretty prints results as CSV line."""
		targets = sorted(results_dict.keys())
		print ",".join(targets)
		print ",".join([str(results_dict[target]) for target in targets])

		def summarize_distribution(y):
		"""Analyzes regression dataset.

deep_chem/utils/preprocess.py

+20 −6

Original line number	Diff line number	Diff line
		@@ -48,12 +48,6 @@ def transform_outputs(dataset, task_transforms, desc_transforms={},
		transforms.update(desc_transforms)
		for task, target in enumerate(endpoints):
		task_transforms = transforms[target]
		print "Task %d has NaNs?" % task
		print np.any(np.isnan(y[:, task]))
		print "Task %d data" % task
		print y[:, task]
		print "Task %d distribution" % task
		summarize_distribution(y[:, task])
		for task_transform in task_transforms:
		if task_transform == "log":
		y[:, task] = np.log(y[:, task])
		@@ -170,6 +164,7 @@ def multitask_to_singletask(dataset):
		# Generate single-task data structures
		labels = dataset.itervalues().next()["labels"]
		sorted_targets = sorted(labels.keys())
		# TODO(rbharath): Replace this with a dictionary comprehension
		singletask = {}
		for target in sorted_targets:
		singletask[target] = {}
		@@ -197,6 +192,8 @@ def train_test_random_split(dataset, frac_train=.8, seed=None):
		----------
		dataset: dict
		A dictionary of type produced by load_datasets.
		frac_train: float
		Proportion of data in train set.
		seed: int (optional)
		Seed to initialize np.random.
		"""
		@@ -211,6 +208,23 @@ def train_test_random_split(dataset, frac_train=.8, seed=None):
		test[key] = dataset[key]
		return train, test

		def train_test_random_split_simple(dataset, frac_train=.8, seed=None):
		"""Splits provided data in train/test splits without separating datasets.

		As opposed to train_test_random_split, this function does not ensure that the
		same compound cannot appear in both train and test (for different targets).

		Parameters
		----------
		dataset: dict
		A dictionary of type produced by load_datasets.
		frac_train: float
		Proportion of data in train set.
		seed: int (optional)
		Seed to initialize np.random.
		"""
		pass

		def train_test_scaffold_split(dataset, frac_train=.8):
		"""Splits provided data into train/test splits by scaffold.

Admin message