Commit cc99dfb5 authored by Bharath Ramsundar's avatar Bharath Ramsundar
Browse files

cleaning up changes

parent 608ef37a

deepchem/models/models.py

+1 −1
Original line number Diff line number Diff line
@@ -77,7 +77,7 @@ class Model(BaseEstimator): 
    raise NotImplementedError(

        "Each model is responsible for its own predict_on_batch method.")



  def restore(self):

  def reload(self):

    """

    Reload trained model from disk.

    """

deepchem/models/sklearn_models/__init__.py

+1 −1
Original line number Diff line number Diff line
@@ -92,7 +92,7 @@ class SklearnModel(Model): 
    """Saves sklearn model to disk using joblib."""

    save_to_disk(self.model_instance, self.get_model_filename(self.model_dir))



  def restore(self):

  def reload(self):

    """Loads sklearn model from joblib file on disk."""

    self.model_instance = load_from_disk(

        Model.get_model_filename(self.model_dir))

deepchem/molnet/run_benchmark.py

+160 −59
Original line number Diff line number Diff line 
# -*- coding: utf-8 -*-

"""

This file provides utilities to run the MoleculeNet benchmark suite.

Created on Mon Mar 06 14:25:40 2017



@author: Zhenqin Wu

"""

import os

import time

import csv

import logging

import numpy as np

import tensorflow as tf

import deepchem
@@ -13,43 +15,6 @@ from deepchem.molnet.run_benchmark_models import benchmark_classification, bench 
from deepchem.molnet.check_availability import CheckFeaturizer, CheckSplit

from deepchem.molnet.preset_hyper_parameters import hps



logger = logging.getLogger(__name__)



# Loading functions available

loading_functions = {

    'bace_c': deepchem.molnet.load_bace_classification,

    'bace_r': deepchem.molnet.load_bace_regression,

    'bbbp': deepchem.molnet.load_bbbp,

    'chembl': deepchem.molnet.load_chembl,

    'clearance': deepchem.molnet.load_clearance,

    'clintox': deepchem.molnet.load_clintox,

    'delaney': deepchem.molnet.load_delaney,

    'factors': deepchem.molnet.load_factors,

    'hiv': deepchem.molnet.load_hiv,

    'hopv': deepchem.molnet.load_hopv,

    'hppb': deepchem.molnet.load_hppb,

    'kaggle': deepchem.molnet.load_kaggle,

    'kinase': deepchem.molnet.load_kinase,

    'lipo': deepchem.molnet.load_lipo,

    'muv': deepchem.molnet.load_muv,

    'nci': deepchem.molnet.load_nci,

    'pcba': deepchem.molnet.load_pcba,

    'pcba_146': deepchem.molnet.load_pcba_146,

    'pcba_2475': deepchem.molnet.load_pcba_2475,

    'pdbbind': deepchem.molnet.load_pdbbind_grid,

    'ppb': deepchem.molnet.load_ppb,

    'qm7': deepchem.molnet.load_qm7_from_mat,

    'qm7b': deepchem.molnet.load_qm7b_from_mat,

    'qm8': deepchem.molnet.load_qm8,

    'qm9': deepchem.molnet.load_qm9,

    'sampl': deepchem.molnet.load_sampl,

    'sider': deepchem.molnet.load_sider,

    'thermosol': deepchem.molnet.load_thermosol,

    'tox21': deepchem.molnet.load_tox21,

    'toxcast': deepchem.molnet.load_toxcast,

    'uv': deepchem.molnet.load_uv,

}





def run_benchmark(datasets,

                  model,
@@ -66,21 +31,16 @@ def run_benchmark(datasets, 
                  test=False,

                  reload=True,

                  seed=123):

  """Run MoleculeNet benchmark suite.



  This is a utility function to help run the MoleculeNet benchmark

  suite on a specified model and a specified dataset.



  Run benchmark test on designated datasets with deepchem(or

  user-defined) model.

  """

  Run benchmark test on designated datasets with deepchem(or user-defined) model



  Parameters

  ----------

  datasets: list of string

      choice of which datasets to use, should be one of: bace_c,

      bace_r, bbbp, chembl, clearance, clintox, delaney, hiv, hopv,

      kaggle, lipo, muv, nci, pcba, pdbbind, ppb, qm7, qm7b, qm8, qm9,

      sampl, sider, tox21, toxcast, uv, factors, kinase

      choice of which datasets to use, should be: bace_c, bace_r, bbbp, chembl,

      clearance, clintox, delaney, hiv, hopv, kaggle, lipo, muv, nci, pcba,

      pdbbind, ppb, qm7, qm7b, qm8, qm9, sampl, sider, tox21, toxcast, uv, factors,

      kinase

  model: string or user-defined model stucture

      choice of which model to use, deepchem provides implementation of

      logistic regression, random forest, multitask network,
@@ -89,10 +49,10 @@ def run_benchmark(datasets, 
  split: string,  optional (default=None)

      choice of splitter function, None = using the default splitter

  metric: string, optional (default=None)

      Choice of evaluation metrics, None = using the default metrics(AUC & R2)

  use_max: bool, (default True)

    Specifies whether to maximize or minimize `metric`.

    maximization(True) or minimization(False)

      choice of evaluation metrics, None = using the default metrics(AUC & R2)

  direction: bool, optional(default=True)

      Optimization direction when doing hyperparameter search

      Maximization(True) or minimization(False)

  featurizer: string or dc.feat.Featurizer,  optional (default=None)

      choice of featurization, None = using the default corresponding to model

      (string only applicable to deepchem models)
@@ -150,12 +110,46 @@ def run_benchmark(datasets, 
    if not split in [None] + CheckSplit[dataset]:

      continue



    logger.info('-------------------------------------')

    logger.info('Benchmark on dataset: %s' % dataset)

    logger.info('-------------------------------------')

    loading_functions = {

        'bace_c': deepchem.molnet.load_bace_classification,

        'bace_r': deepchem.molnet.load_bace_regression,

        'bbbp': deepchem.molnet.load_bbbp,

        'chembl': deepchem.molnet.load_chembl,

        'clearance': deepchem.molnet.load_clearance,

        'clintox': deepchem.molnet.load_clintox,

        'delaney': deepchem.molnet.load_delaney,

        'factors': deepchem.molnet.load_factors,

        'hiv': deepchem.molnet.load_hiv,

        'hopv': deepchem.molnet.load_hopv,

        'hppb': deepchem.molnet.load_hppb,

        'kaggle': deepchem.molnet.load_kaggle,

        'kinase': deepchem.molnet.load_kinase,

        'lipo': deepchem.molnet.load_lipo,

        'muv': deepchem.molnet.load_muv,

        'nci': deepchem.molnet.load_nci,

        'pcba': deepchem.molnet.load_pcba,

        'pcba_146': deepchem.molnet.load_pcba_146,

        'pcba_2475': deepchem.molnet.load_pcba_2475,

        'pdbbind': deepchem.molnet.load_pdbbind_grid,

        'ppb': deepchem.molnet.load_ppb,

        'qm7': deepchem.molnet.load_qm7_from_mat,

        'qm7b': deepchem.molnet.load_qm7b_from_mat,

        'qm8': deepchem.molnet.load_qm8,

        'qm9': deepchem.molnet.load_qm9,

        'sampl': deepchem.molnet.load_sampl,

        'sider': deepchem.molnet.load_sider,

        'thermosol': deepchem.molnet.load_thermosol,

        'tox21': deepchem.molnet.load_tox21,

        'toxcast': deepchem.molnet.load_toxcast,

        'uv': deepchem.molnet.load_uv,

    }



    print('-------------------------------------')

    print('Benchmark on dataset: %s' % dataset)

    print('-------------------------------------')

    # loading datasets

    if split is not None:

      logger.info('Splitting function: %s' % split)

      print('Splitting function: %s' % split)

      tasks, all_dataset, transformers = loading_functions[dataset](

          featurizer=featurizer, split=split, reload=reload)

    else:
@@ -179,7 +173,8 @@ def run_benchmark(datasets, 
          valid_dataset,

          transformers,

          metric,

          use_max=use_max,

          direction=direction,

          n_features=n_features,

          n_tasks=len(tasks),

          max_iter=max_iter,

          search_range=search_range)
@@ -192,6 +187,7 @@ def run_benchmark(datasets, 
            test_dataset,

            tasks,

            transformers,

            n_features,

            metric,

            model,

            test=test,
@@ -239,3 +235,108 @@ def run_benchmark(datasets, 
    if hyper_param_search:

      with open(os.path.join(out_path, dataset + model + '.pkl'), 'w') as f:

        pickle.dump(hyper_parameters, f)





#

# Note by @XericZephyr. Reason why I spun off this function:

#   1. Some model needs dataset information.

#   2. It offers us possibility to **cache** the dataset

#      if the featurizer runs very slow, e.g., GraphConv.

#   2+. The cache can even happen at Travis CI to accelerate

#       CI testing.

#

def load_dataset(dataset, featurizer, split='random'):

  """

  Load specific dataset for benchmark.



  Parameters

  ----------

  dataset: string

      choice of which datasets to use, should be: tox21, muv, sider,

      toxcast, pcba, delaney, factors, hiv, hopv, kaggle, kinase, nci,

      clintox, hiv, pcba_128, pcba_146, pdbbind, chembl, qm7, qm7b, qm9,

      sampl, uv

  featurizer: string or dc.feat.Featurizer.

      choice of featurization.

  split: string,  optional (default=None)

      choice of splitter function, None = using the default splitter

  """

  dataset_loading_functions = {

      'bace_c': deepchem.molnet.load_bace_classification,

      'bace_r': deepchem.molnet.load_bace_regression,

      'bbbp': deepchem.molnet.load_bbbp,

      'chembl': deepchem.molnet.load_chembl,

      'clearance': deepchem.molnet.load_clearance,

      'clintox': deepchem.molnet.load_clintox,

      'delaney': deepchem.molnet.load_delaney,

      'factors': deepchem.molnet.load_factors,

      'hiv': deepchem.molnet.load_hiv,

      'hopv': deepchem.molnet.load_hopv,

      'hppb': deepchem.molnet.load_hppb,

      'kaggle': deepchem.molnet.load_kaggle,

      'kinase': deepchem.molnet.load_kinase,

      'lipo': deepchem.molnet.load_lipo,

      'muv': deepchem.molnet.load_muv,

      'nci': deepchem.molnet.load_nci,

      'pcba': deepchem.molnet.load_pcba,

      'pcba_128': deepchem.molnet.load_pcba_128,

      'pcba_146': deepchem.molnet.load_pcba_146,

      'pcba_2475': deepchem.molnet.load_pcba_2475,

      'pdbbind': deepchem.molnet.load_pdbbind_grid,

      'ppb': deepchem.molnet.load_ppb,

      'qm7': deepchem.molnet.load_qm7_from_mat,

      'qm7b': deepchem.molnet.load_qm7b_from_mat,

      'qm8': deepchem.molnet.load_qm8,

      'qm9': deepchem.molnet.load_qm9,

      'sampl': deepchem.molnet.load_sampl,

      'sider': deepchem.molnet.load_sider,

      'thermosol': deepchem.molnet.load_thermosol,

      'tox21': deepchem.molnet.load_tox21,

      'toxcast': deepchem.molnet.load_toxcast,

      'uv': deepchem.molnet.load_uv

  }

  print('-------------------------------------')

  print('Loading dataset: %s' % dataset)

  print('-------------------------------------')

  # loading datasets

  if split is not None:

    print('Splitting function: %s' % split)

  tasks, all_dataset, transformers = dataset_loading_functions[dataset](

      featurizer=featurizer, split=split)

  return tasks, all_dataset, transformers





def benchmark_model(model, all_dataset, transformers, metric, test=False):

  """

  Benchmark custom model.



  model: user-defined model stucture

    For user define model, it should include function: fit, evaluate.



  all_dataset: (train, test, val) data tuple.

    Returned by `load_dataset` function.



  transformers



  metric: string

    choice of evaluation metrics.





  """

  time_start_fitting = time.time()

  train_score = .0

  valid_score = .0

  test_score = .0



  train_dataset, valid_dataset, test_dataset = all_dataset



  model.fit(train_dataset)

  train_score = model.evaluate(train_dataset, metric, transformers)

  valid_score = model.evaluate(valid_dataset, metric, transformers)

  if test:

    test_score = model.evaluate(test_dataset, metric, transformers)



  time_finish_fitting = time.time()

  time_for_running = time_finish_fitting - time_start_fitting



  return train_score, valid_score, test_score, time_for_running

deepchem/molnet/run_benchmark_models.py

+2 −0
Original line number Diff line number Diff line 
#!/usr/bin/env python2

# -*- coding: utf-8 -*-

"""

Created on Mon Mar  6 23:41:26 2017

docs/featurizers.rst

+0 −19
Original line number Diff line number Diff line
@@ -22,25 +22,6 @@ nevertheless, deep learning systems can't simply chew up raw files. 
For this reason, :code:`deepchem` provides an extensive collection of

featurization methods which we will review on this page.



Featurizer-Model Matchups

-------------------------



If you're using DeepChem in practical applications, you probably want

to use a given model on some dataset. Your first question when you try

to do this will probably be which featurizer should I use?



+------------+--------------------------+-----------+

| Model      | Acceptable Featurizers   | Header 3  |

+============+==========================+===========+

| body row 1 | column 2   | column 3  |

+------------+------------+-----------+

| body row 2 | Cells may span columns.|

+------------+------------+-----------+

| body row 3 | Cells may  | - Cells   |

+------------+ span rows. | - contain |

| body row 4 |            | - blocks. |

+------------+------------+-----------+



Featurizer

----------

CRA Git | Maintained and supported by SUSTech CRA and CCSE