Added singletask/multitask support.

"""

Code for processing the Google vs-datasets using scikit-learn.

Code for processing datasets using scikit-learn.

"""

import numpy as np

from deep_chem.utils.load import load_and_transform_dataset

  Parameters

  ----------

  paths: list 

    List of paths to Google vs datasets. 

    List of paths to datasets. 

  modeltype: String

    A string describing the model to be trained. Options are RandomForest,

  splittype: string

      model = LassoLarsCV(max_iter=2000, n_jobs=-1) 

    elif modeltype == "elastic_net":

      model = ElasticNetCV(max_iter=2000, n_jobs=-1) 

    elif modeltype == "svr_rbf":

      model = SVR(kernel="rbf") 

    elif modeltype == "svr_poly":

      model = SVR(kernel="poly") 

    elif modeltype == "svr_linear":

      model = SVR(kernel="linear") 

    else:

      raise ValueError("Invalid model type provided.")

    model.fit(X_train, y_train.ravel())

def parse_args(input_args=None):

  """Parse command-line arguments."""

  parser = argparse.ArgumentParser()

  parser.add_argument('--assay', required=1,

                      help='Assay ID.')

  parser.add_argument('--dataset', required=1, choices=['muv', 'pcba', 'dude', 'pfizer'],

  parser.add_argument('--dataset', required=1, nargs="+",

                      choices=['muv', 'pcba', 'dude', 'pfizer'],

                      help='Name of dataset to process.')

  parser.add_argument('--model', required=1, nargs="+",

                      choices=["logistic", "rf_classifier", "single_task_deep_network"])

  parser.add_argument('--model', required=1,

                      choices=["logistic", "rf_classifier", "rf_regressor",

                      "linear", "ridge", "lasso", "lasso_lars", "elastic_net",

                      "singletask_deep_network", "multitask_deep_network"])

  parser.add_argument("--splittype", type=str, default="scaffold",

                       choices=["scaffold", "random"],

                       help="Type of cross-validation data-splitting.")

  parser.add_argument("--n-hidden", type=int, default=500,

                      help="Number of hidden neurons for NN models.")

  parser.add_argument("--learning-rate", type=float, default=0.01,

                  help="Learning rate for NN models.")

  parser.add_argument("--dropout", type=float, default=0.5,

                  help="Learning rate for NN models.")

  parser.add_argument("--n-epochs", type=int, default=50,

                  help="Number of epochs for NN models.")

  parser.add_argument("--batchsize", type=int, default=32,

                  help="Number of examples per minibatch for NN models.")

  parser.add_argument("--decay", type=float, default=1e-4,

                  help="Learning rate decay for NN models.")

  parser.add_argument("--validation-split", type=float, default=0.0,

                  help="Percent of training data to use for validation.")

  return parser.parse_args(input_args)

def main():

  args = parse_args()

  if args.dataset == "muv":

    path = "/home/rbharath/vs-datasets/muv"

  elif args.dataset == "pcba":

    path = "/home/rbharath/vs-datasets/pcba"

  elif args.dataset == "dude":

    path = "/home/rbharath/vs-datasets/dude"

  # TODO(rbharath): The pfizer dataset is private. Remove this before the

  # public release of the code.

  elif args.dataset == "pfizer":

    path = "/home/rbharath/private-datasets/pfizer"

  paths = {}

  if "muv" in args.dataset:

    paths["muv"] = "/home/rbharath/vs-datasets/muv"

  elif "pcba" in args.dataset:

    paths["pcba"] = "/home/rbharath/vs-datasets/pcba"

  elif "dude" in args.dataset:

    paths["dude"]= "/home/rbharath/vs-datasets/dude"

  # TODO(rbharath): The pfizer dataset is currently private. Remove this before

  # the public release of the code.

  elif "pfizer" in args.dataset:

    paths["pfizer"] = "/home/rbharath/private-datasets/pfizer"

  task_types, task_transforms = get_default_task_types_and_transforms(

    {args.dataset: path})

  task_types, task_transforms = get_default_task_types_and_transforms(paths)

  desc_transforms = get_default_descriptor_transforms()

  if len(args.model) == 1:

    model = args.model[0]

    fit_singletask_models([path], model, task_types,

  if args.model == "singletask_deep_network":

    fit_singletask_mlp(paths.values(), task_types, task_transforms,

      desc_transforms, splittype=args.splittype, add_descriptors=False,

      n_hidden=args.n_hidden, learning_rate=args.learning_rate,

      dropout=args.dropout, nb_epoch=args.n_epochs, decay=args.decay,

      validation_split=args.validation_split)

  elif args.model == "multitask_deep_network":

    fit_multitask_mlp(paths.values(), task_types, task_transforms,

      desc_transforms, splittype=args.splittype, add_descriptors=False,

      n_hidden=args.n_hidden, learning_rate = args.learning_rate, dropout = args.dropout,

      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")

  #fit_multitask_mlp([muv_path, pfizer_path], task_types, task_transforms,

  #  desc_transforms, splittype="scaffold", add_descriptors=False,

  #  desc_weight=0.1, n_hidden=500, learning_rate = .01, dropout = .5,

  #  nb_epoch=50, decay=1e-4, validation_split=0.01)

  #fit_multitask_mlp([muv_path, pfizer_path], task_types, task_transforms,

  #  desc_transforms, splittype="scaffold", add_descriptors=False, n_hidden=500,

  #  nb_epoch=40, learning_rate=0.01, decay=1e-4, dropout = .5)

  #fit_multitask_mlp([dude_path], task_types, task_transforms,

  #  desc_transforms, splittype="scaffold", add_descriptors=False, n_hidden=500,

  #  nb_epoch=40, learning_rate=0.01, decay=1e-4, dropout = .5)

  #fit_multitask_mlp([muv_path], task_types, task_transforms, desc_transforms,

  #  splittype="scaffold", add_descriptors=False, n_hidden=500,

  #  learning_rate=.01, dropout=.5, nb_epoch=30, decay=1e-4)

  #fit_singletask_mlp([muv_path], task_types, task_transforms, desc_transforms,

  #  splittype="scaffold", add_descriptors=False, n_hidden=500,

  #  learning_rate=.01, dropout=.5, nb_epoch=30, decay=1e-4)

if __name__ == "__main__":

  main()