Commit 17f082bb authored by Bharath Ramsundar's avatar Bharath Ramsundar
Browse files

Changes for experiments

parent 724a4fac

deepchem/models/tf_keras_models/support_classifier.py

+1 −1
Original line number Diff line number Diff line
@@ -340,7 +340,7 @@ class SupportGraphClassifier(Model): 
    feed_dict = self.construct_feed_dict(padded_test_batch, support)

    # Get scores

    pred, scores = self.sess.run([self.pred_op, self.scores_op], feed_dict=feed_dict)

    y_pred_batch = np.round(scores)

    y_pred_batch = np.round(pred)

    ########################################################### DEBUG

    # Remove padded elements

    y_pred_batch = y_pred_batch[:n_samples]

examples/low_data/datasets.py

+66 −0
Original line number Diff line number Diff line
@@ -12,6 +12,10 @@ import tempfile 
import numpy as np

import deepchem as dc



def to_numpy_dataset(dataset):

  """Converts dataset to numpy dataset."""

  return dc.data.NumpyDataset(dataset.X, dataset.y, dataset.w, dataset.ids)



def load_tox21_ecfp(num_train=7200):

  """Load Tox21 datasets. Does not do train/test split"""

  # Set some global variables up top
@@ -128,3 +132,65 @@ def load_muv_convmol(): 
      dataset = transformer.transform(dataset)



  return MUV_tasks, dataset, transformers



def load_sider_ecfp():

  """Load SIDER datasets. Does not do train/test split"""

  # Featurize SIDER dataset

  print("About to featurize SIDER dataset.")

  current_dir = os.path.dirname(os.path.realpath(__file__))

  dataset_file = os.path.join(

      current_dir, "../sider/sider.csv.gz")

  featurizer = dc.feat.CircularFingerprint(size=1024)



  dataset = dc.utils.save.load_from_disk(dataset_file)

  SIDER_tasks = dataset.columns.values[1:].tolist()

  print("SIDER tasks: %s" % str(SIDER_tasks))

  print("%d tasks in total" % len(SIDER_tasks))





  loader = dc.load.DataLoader(tasks=SIDER_tasks,

                      smiles_field="smiles",

                      featurizer=featurizer,

                      verbosity="high")

  dataset = loader.featurize(dataset_file)

  print("%d datapoints in SIDER dataset" % len(dataset))



  # Initialize transformers

  transformers = [

      dc.trans.BalancingTransformer(transform_w=True, dataset=dataset)]

  print("About to transform data")

  for transformer in transformers:

    dataset = transformer.transform(dataset)



  return SIDER_tasks, dataset, transformers



def load_sider_convmol():

  """Load SIDER datasets. Does not do train/test split"""

  # Featurize SIDER dataset

  print("About to featurize SIDER dataset.")

  current_dir = os.path.dirname(os.path.realpath(__file__))

  dataset_file = os.path.join(

      current_dir, "../sider/sider.csv.gz")

  featurizer = dc.feat.ConvMolFeaturizer()



  dataset = dc.utils.save.load_from_disk(dataset_file)

  SIDER_tasks = dataset.columns.values[1:].tolist()

  print("SIDER tasks: %s" % str(SIDER_tasks))

  print("%d tasks in total" % len(SIDER_tasks))





  loader = dc.load.DataLoader(tasks=SIDER_tasks,

                      smiles_field="smiles",

                      featurizer=featurizer,

                      verbosity="high")

  dataset = loader.featurize(dataset_file, debug=True)

  print("%d datapoints in SIDER dataset" % len(dataset))



  # Initialize transformers

  transformers = [

      dc.trans.BalancingTransformer(transform_w=True, dataset=dataset)]

  print("About to transform data")

  for transformer in transformers:

    dataset = transformer.transform(dataset)



  return SIDER_tasks, dataset, transformers

examples/low_data/muv_attn_one_fold.py

+1 −1
Original line number Diff line number Diff line
@@ -17,7 +17,7 @@ K = 4 
max_depth = 3

# number positive/negative ligands

n_pos = 1

n_neg = 5

n_neg = 1

# Set batch sizes for network

test_batch_size = 128

support_batch_size = n_pos + n_neg

examples/low_data/muv_attn_one_fold_no_train.py

0 → 100644
+83 −0
Original line number Diff line number Diff line 
"""

Train low-data attn models on MUV. Test last fold only.

"""

from __future__ import print_function

from __future__ import division

from __future__ import unicode_literals



import tempfile

import numpy as np

import deepchem as dc

import tensorflow as tf

from datasets import load_muv_convmol



# Number of folds for split 

K = 4 

# Depth of attention module

max_depth = 3

# number positive/negative ligands

n_pos = 10

n_neg = 10

# Set batch sizes for network

test_batch_size = 128

support_batch_size = n_pos + n_neg

nb_epochs = 1

n_train_trials = 2000

n_eval_trials = 20

learning_rate = 1e-4

log_every_n_samples = 50

# Number of features on conv-mols

n_feat = 71



muv_tasks, dataset, transformers = load_muv_convmol()



# Define metric

metric = dc.metrics.Metric(

    dc.metrics.roc_auc_score, verbosity="high", mode="classification")



task_splitter = dc.splits.TaskSplitter()

fold_datasets = task_splitter.k_fold_split(dataset, K)



train_folds = fold_datasets[:-1] 

train_dataset = dc.splits.merge_fold_datasets(train_folds)

test_dataset = fold_datasets[-1]



# Train support model on train

support_model = dc.nn.SequentialSupportGraph(n_feat)



# Add layers

support_model.add(dc.nn.GraphConv(64, activation='relu'))

support_model.add(dc.nn.GraphPool())

support_model.add(dc.nn.GraphConv(128, activation='relu'))

support_model.add(dc.nn.GraphPool())

support_model.add(dc.nn.GraphConv(64, activation='relu'))

support_model.add(dc.nn.GraphPool())

support_model.add(dc.nn.Dense(128, activation='tanh'))



support_model.add_test(dc.nn.GraphGather(test_batch_size, activation='tanh'))

support_model.add_support(dc.nn.GraphGather(support_batch_size, activation='tanh'))



# Apply an attention lstm layer

support_model.join(dc.nn.AttnLSTMEmbedding(

    test_batch_size, support_batch_size, max_depth))



with tf.Session() as sess:

  model = dc.models.SupportGraphClassifier(

    sess, support_model, test_batch_size=test_batch_size,

    support_batch_size=support_batch_size, learning_rate=learning_rate,

    verbosity="high")



  # Turning off training to investigate

  ############################################################# DEBUG

  #print("FIT")

  ############################################################# DEBUG

  #model.fit(train_dataset, nb_epochs=nb_epochs, 

  #          n_episodes_per_epoch=n_train_trials,

  #          n_pos=n_pos, n_neg=n_neg, log_every_n_samples=log_every_n_samples)

  ############################################################ DEBUG

  print("EVAL")

  ############################################################ DEBUG

  scores = model.evaluate(

      test_dataset, metric, n_pos, n_neg, n_trials=n_eval_trials)

  print("Scores on evaluation dataset")

  print(scores)

examples/low_data/muv_res_one_fold.py

+1 −1
Original line number Diff line number Diff line
@@ -17,7 +17,7 @@ K = 4 
max_depth = 3

# num positive/negative ligands

n_pos = 1

n_neg = 5

n_neg = 1

# Set batch sizes for network

test_batch_size = 128

support_batch_size = n_pos + n_neg

CRA Git | Maintained and supported by SUSTech CRA and CCSE