import os import numpy as np import deepchem as dc from deepchem.molnet import load_pdbbind_grid pdbbind_tasks, pdbbind_datasets, transformers = load_pdbbind_grid(split='random', subset='full') train_data, valid_data, test_data = pdbbind_datasets metric = dc.metrics.Metric(dc.metrics.pearson_r2_score) model = dc.models.MultitaskRegressor(len(pdbbind_tasks), train_data.X.shape[1],dropouts=[.25],learning_rate=0.0003,weight_init_stddevs=[.1],batch_size=64, model_dir='pdbbind_test') model.fit(train_data, nb_epoch=100) train_score = model.evaluate(train_data, [metric], transformers) valid_score = model.evaluate(valid_data, [metric], transformers) feature = dc.feat.RdkitGridFeaturizer(voxel_width=16.0,feature_types=["ecfp", "splif", "hbond", "salt_bridge"],ecfp_power=9,splif_power=9,flatten=True) grid = feature.featurize_complexes(['ligand2.sdf'], ['3lpt.pdb']) print(grid) # reload = dc.models.MultitaskRegressor(len(pdbbind_tasks), train_data.X.shape[1],dropouts=[.25],learning_rate=0.0003,weight_init_stddevs=[.1],batch_size=64, model_dir='pdbbind_test') # print(reload.get_checkpoints()) # reload.load_from_pretrained(reload, model_dir='pdbbind_test') # reload.evaluate(train_data, [metric], transformers) # reload.evaluate(valid_data, [metric], transformers)
__copyright__ = "Copyright 2016, Stanford University"
__license__ = "MIT"
import os
import numpy as np
import tensorflow as tf
# For stable runs
np.random.seed(123)
tf.set_random_seed(123)
import deepchem as dc
from deepchem.molnet import load_pdbbind_grid
split = "random"
subset = "full"
pdbbind_tasks, pdbbind_datasets, transformers = load_pdbbind_grid(
split=split, subset=subset)
train_dataset, valid_dataset, test_dataset = pdbbind_datasets
metric = dc.metrics.Metric(dc.metrics.pearson_r2_score)
current_dir = os.path.dirname(os.path.realpath(__file__))
model_dir = os.path.join(current_dir, "%s_%s_DNN" % (split, subset))
n_features = train_dataset.X.shape[1]
model = dc.models.MultitaskRegressor(len(pdbbind_tasks),
n_features,
model_dir=model_dir,
dropouts=[.25],
learning_rate=0.0003,
weight_init_stddevs=[.1],
batch_size=64)
__author__ = "Bharath Ramsundar"
__copyright__ = "Copyright 2016, Stanford University"
__license__ = "MIT"
import os
import deepchem as dc
import numpy as np
from sklearn.ensemble import RandomForestRegressor
from deepchem.molnet import load_pdbbind_grid
# For stable runs
np.random.seed(123)
split = "random"
subset = "full"
pdbbind_tasks, pdbbind_datasets, transformers = load_pdbbind_grid(
split=split, subset=subset)
train_dataset, valid_dataset, test_dataset = pdbbind_datasets
metric = dc.metrics.Metric(dc.metrics.pearson_r2_score)
current_dir = os.path.dirname(os.path.realpath(__file__))
model_dir = os.path.join(current_dir, "%s_%s_RF" % (split, subset))
sklearn_model = RandomForestRegressor(n_estimators=500)
model = dc.models.SklearnModel(sklearn_model, model_dir=model_dir)
# Fit trained model
print("Fitting model on train dataset")
model.fit(train_dataset)
model.save()