Abstract class for all models. Parameters: ----------- model_instance: object Wrapper around ScikitLearn/Keras/Tensorflow model obje

Constructs the computational graph. This function constructs the computational graph for the model. It relies subclassed methods (build/cost)

Reward of a state. Returns: Float. QED of the current state.

Add a single graph Parameters ---------- g : tg.data.Data Molecular graph

Creates a list of valences corresponding to atom_types. Note that this is not a count of valence electrons, but a count of the maximum number of

Constructs the graph architecture as specified in its config. This method creates the following Placeholders: mol_features: Molecule des

Constructs the graph architecture as specified in its config. This method creates the following Placeholders: mol_features: Molecule descri

clone() method to trick Theano

Returns whether mol contains the given scaffold. NOTE: This is more advanced than simply computing scaffold equality (i.e. scaffold(mol_a) == sca

Returns ------- str File path of the next data batch.

Method to represent attributed graph as a giant tensor The tensor is N_node x N_node x N_attributes. For a given node, A_i,i,: is a vect

Stop the training if there is no non-trivial progress in k steps @param val_acc_history: a list contains all the historical validation acc @par

Returns 2D array where (#, :) contains attributes of edge #

Fits a model on data in a Dataset object.

Fit the model. Parameters ---------- dataset: dc.data.Dataset Dataset object holding training data nb_epoch: 10 Number

Updates existing model with new information.

Take in and process masked src and target sequences.

Pass the input (and mask) through each layer in turn.

Apply residual connection to any sublayer with the same size.

Follow Figure 1 (left) for connections.

Implements Figure 2

TODO: Tutaj mozna jeszcze wrzucic jakas nieliniowosc na koniec, bo gdy N_dense = 1, to mamy zwykle mnozenie macierzy

Get Morgan Fingerprint of a specific SMILES string. Args: smiles: String. The SMILES string of the molecule. hparams: tf.contrib.training.H

Get number of tasks.

Computes the Bemis-Murcko scaffold for a molecule. Args: mol: RDKit Mol. Returns: String scaffold SMILES.

Currently models can only be classifiers or regressors.

Load dataset from a csv file Parameters ---------- path : str Path to the csv data file. label_col : list of str Column names for

Load dataset from a csv file Parameters ---------- path : str Path to the sdf data file. label_col : list of str Column names for

Customized function for DataLoader that dynamically pads the batch so that all data have the same length

Returns 2D array where (#, :) contains attributes of node #

Calculates the penalized logP of a molecule. Refactored from https://github.com/wengong-jin/icml18-jtnn/blob/master/bo/run_bo.py See Junction T

Uses self to make predictions on provided Dataset object. Returns: y_pred: numpy ndarray of shape (n_samples,)

Makes predictions on given batch of new data. Parameters ---------- X: np.ndarray Features

Returns ------- list File paths of saved data batches.

Reload trained model from disk.

Reorder the data batches according to new indices Usually used after shuffle Parameters ---------- indices : list, optional

Test that classification init can happen.

Simple test that initializes and fits a MAT.

Test conversion of molecule to graph.

Test that regression init can happen.

Draws the molecule of the state. Args: state: String, Chem.Mol, or Chem.RWMol. If string is prov ided, it is considered as the SMIL