Commit 79176a19 authored by miaecle's avatar miaecle
Browse files

refining DTNN

parent c08d9e1d

deepchem/models/__init__.py

+0 −1
Original line number Diff line number Diff line
@@ -10,7 +10,6 @@ from deepchem.models.sklearn_models import SklearnModel 
from deepchem.models.xgboost_models import XGBoostModel

from deepchem.models.tf_new_models.multitask_classifier import MultitaskGraphClassifier

from deepchem.models.tf_new_models.multitask_regressor import MultitaskGraphRegressor

from deepchem.models.tf_new_models.DTNN_regressor import DTNNGraphRegressor



from deepchem.models.tf_new_models.support_classifier import SupportGraphClassifier

from deepchem.models.multitask import SingletaskToMultitask

deepchem/models/tf_new_models/graph_topology.py

+14 −18
Original line number Diff line number Diff line
@@ -169,6 +169,10 @@ class DTNNGraphTopology(GraphTopology): 
    self.n_distance = n_distance

    self.distance_min = distance_min

    self.distance_max = distance_max

    self.step_size = (distance_max - distance_min) / n_distance

    self.steps = np.array(

        [distance_min + i * self.step_size for i in range(n_distance)])

    self.steps = np.expand_dims(self.steps, 0)



    self.atom_number_placeholder = tf.placeholder(

        dtype='int32', shape=(None,), name=self.name + '_atom_number')
@@ -179,9 +183,9 @@ class DTNNGraphTopology(GraphTopology): 
    self.atom_membership_placeholder = tf.placeholder(

        dtype='int32', shape=(None,), name=self.name + '_atom_membership')

    self.distance_membership_i_placeholder = tf.placeholder(

        dtype='int32', shape=(None,), name=self.name + '_distance_membership')

        dtype='int32', shape=(None,), name=self.name + '_distance_membership_i')

    self.distance_membership_j_placeholder = tf.placeholder(

        dtype='int32', shape=(None,), name=self.name + '_distance_membership')

        dtype='int32', shape=(None,), name=self.name + '_distance_membership_j')



    # Define the list of tensors to be used as topology

    self.topology = [
@@ -222,19 +226,17 @@ class DTNNGraphTopology(GraphTopology): 
            np.power(2 * np.diag(batch[i, :num_atoms[i], :num_atoms[i]]), 1 /

                     2.4)).astype(int) for i in range(len(num_atoms))

    ]



    distance = []

    atom_membership = []

    distance_membership_i = []

    distance_membership_j = []

    start = 0

    for im, molecule in enumerate(atom_number):

      distance_matrix = batch[im, :num_atoms[im], :num_atoms[im]] / np.outer(

          molecule, molecule)

      np.fill_diagonal(distance_matrix, 0)

      distance_matrix = np.expand_dims(distance_matrix.flatten(), 1)

      distance_matrix = self.gauss_expand(distance_matrix, self.n_distance,

                                          self.distance_min, self.distance_max)

      distance.append(distance_matrix)

      distance_matrix = np.outer(

          molecule, molecule) / batch[im, :num_atoms[im], :num_atoms[im]]

      np.fill_diagonal(distance_matrix, -100)

      distance.append(np.expand_dims(distance_matrix.flatten(), 1))

      atom_membership.append([im] * num_atoms[im])

      membership = np.array([np.arange(num_atoms[im])] * num_atoms[im])

      membership_i = membership.flatten(order='F')
@@ -243,7 +245,9 @@ class DTNNGraphTopology(GraphTopology): 
      distance_membership_j.append(membership_j + start)

      start = start + num_atoms[im]

    atom_number = np.concatenate(atom_number)

    distance = np.concatenate(distance)

    distance = np.concatenate(distance, 0)

    distance = np.exp(-np.square(distance - self.steps) /

                      (2 * self.step_size**2))

    distance_membership_i = np.concatenate(distance_membership_i)

    distance_membership_j = np.concatenate(distance_membership_j)

    atom_membership = np.concatenate(atom_membership)
@@ -257,14 +261,6 @@ class DTNNGraphTopology(GraphTopology): 
    }

    return dict_DTNN



  @staticmethod

  def gauss_expand(distance, n_distance, distance_min, distance_max):

    step_size = (distance_max - distance_min) / n_distance

    steps = np.array([distance_min + i * step_size for i in range(n_distance)])

    steps = np.expand_dims(steps, 0)

    distance_vector = np.exp(-np.square(distance - steps) / (2 * step_size**2))

    return distance_vector





class DAGGraphTopology(GraphTopology):

  """GraphTopology for DAG models

deepchem/molnet/check_availability.py

+8 −8
Original line number Diff line number Diff line
@@ -142,14 +142,14 @@ CheckFeaturizer = { 
    ('kaggle', 'rf_regression'): [None, 14293],

    ('pdbbind', 'tf_regression'): ['grid', 2052],

    ('pdbbind', 'rf_regression'): ['grid', 2052],

    ('qm7', 'tf_regression_ft'): [None, [23, 23]],

    ('qm7', 'dtnn'): [None, [23, 23]],

    ('qm7b', 'tf_regression_ft'): [None, [23, 23]],

    ('qm7b', 'dtnn'): [None, [23, 23]],

    ('qm8', 'tf_regression_ft'): [None, [26, 26]],

    ('qm8', 'dtnn'): [None, [26, 26]],

    ('qm9', 'tf_regression_ft'): [None, [29, 29]],

    ('qm9', 'dtnn'): [None, [29, 29]]

    ('qm7', 'tf_regression_ft'): ['CoulombMatrix', [23, 23]],

    ('qm7', 'dtnn'): ['CoulombMatrix', [23, 23]],

    ('qm7b', 'tf_regression_ft'): ['CoulombMatrix', [23, 23]],

    ('qm7b', 'dtnn'): ['CoulombMatrix', [23, 23]],

    ('qm8', 'tf_regression_ft'): ['CoulombMatrix', [26, 26]],

    ('qm8', 'dtnn'): ['CoulombMatrix', [26, 26]],

    ('qm9', 'tf_regression_ft'): ['CoulombMatrix', [29, 29]],

    ('qm9', 'dtnn'): ['CoulombMatrix', [29, 29]]

}



CheckSplit = {

deepchem/molnet/load_function/bace_datasets.py

+4 −4
Original line number Diff line number Diff line
@@ -10,7 +10,7 @@ import deepchem 
from deepchem.molnet.load_function.bace_features import bace_user_specified_features





def load_bace_regression(featurizer=None, split='random', reload=True):

def load_bace_regression(featurizer='ECFP', split='random', reload=True):

  """Load bace datasets."""

  # Featurize bace dataset

  print("About to featurize bace dataset.")
@@ -44,7 +44,7 @@ def load_bace_regression(featurizer=None, split='random', reload=True): 
    featurizer = deepchem.feat.WeaveFeaturizer()

  elif featurizer == 'Raw':

    featurizer = deepchem.feat.RawFeaturizer()

  elif featurizer == None:

  elif featurizer == 'UserDefined':

    featurizer = deepchem.feat.UserDefinedFeaturizer(

        bace_user_specified_features)
@@ -76,7 +76,7 @@ def load_bace_regression(featurizer=None, split='random', reload=True): 
  return bace_tasks, (train, valid, test), transformers





def load_bace_classification(featurizer=None, split='random', reload=True):

def load_bace_classification(featurizer='ECFP', split='random', reload=True):

  """Load bace datasets."""

  # Featurize bace dataset

  print("About to featurize bace dataset.")
@@ -110,7 +110,7 @@ def load_bace_classification(featurizer=None, split='random', reload=True): 
    featurizer = deepchem.feat.WeaveFeaturizer()

  elif featurizer == 'Raw':

    featurizer = deepchem.feat.RawFeaturizer()

  elif featurizer == None:

  elif featurizer == 'UserDefined':

    featurizer = deepchem.feat.UserDefinedFeaturizer(

        bace_user_specified_features)

deepchem/molnet/load_function/qm7_datasets.py

+4 −4
Original line number Diff line number Diff line
@@ -11,7 +11,7 @@ import deepchem 
import scipy.io





def load_qm7_from_mat(featurizer=None, split='stratified', reload=True):

def load_qm7_from_mat(featurizer='CoulombMatrix', split='stratified', reload=True):

  if "DEEPCHEM_DATA_DIR" in os.environ:

    data_dir = os.environ["DEEPCHEM_DATA_DIR"]

  else:
@@ -57,7 +57,7 @@ def load_qm7_from_mat(featurizer=None, split='stratified', reload=True): 
  return qm7_tasks, (train_dataset, valid_dataset, test_dataset), transformers





def load_qm7b_from_mat(featurizer=None, split='stratified', reload=True):

def load_qm7b_from_mat(featurizer='CoulombMatrix', split='stratified', reload=True):

  if "DEEPCHEM_DATA_DIR" in os.environ:

    data_dir = os.environ["DEEPCHEM_DATA_DIR"]

  else:
@@ -100,7 +100,7 @@ def load_qm7b_from_mat(featurizer=None, split='stratified', reload=True): 
  return qm7_tasks, (train_dataset, valid_dataset, test_dataset), transformers





def load_qm7(featurizer=None, split='random', reload=True):

def load_qm7(featurizer='CoulombMatrix', split='random', reload=True):

  """Load qm7 datasets."""

  # Featurize qm7 dataset

  print("About to featurize qm7 dataset.")
@@ -120,7 +120,7 @@ def load_qm7(featurizer=None, split='random', reload=True): 
              data_dir)



  qm7_tasks = ["u0_atom"]

  if featurizer is None:

  if featurizer == 'CoulombMatrix':

    featurizer = deepchem.feat.CoulombMatrixEig(23)

  loader = deepchem.data.SDFLoader(

      tasks=qm7_tasks,

CRA Git | Maintained and supported by SUSTech CRA and CCSE