Multi-objective de novo drug design with conditional graph generative model

Abstract

Recently, deep generative models have revealed itself as a promising way of performing de novo molecule design. However, previous research has focused mainly on generating SMILES strings instead of molecular graphs. Although available, current graph generative models are are often too general and computationally expensive. In this work, a new de novo molecular design framework is proposed based on a type of sequential graph generators that do not use atom level recurrent units. Compared with previous graph generative models, the proposed method is much more tuned for molecule generation and has been scaled up to cover significantly larger molecules in the ChEMBL database. It is shown that the graph-based model outperforms SMILES based models in a variety of metrics, especially in the rate of valid outputs. For the application of drug design tasks, conditional graph generative model is employed. This method offers highe flexibility and is suitable for generation based on multiple objectives. The results have demonstrated that this approach can be effectively applied to solve several drug design problems, including the generation of compounds containing a given scaffold, compounds with specific drug-likeness and synthetic accessibility requirements, as well as dual inhibitors against JNK3 and GSK-3β. Electronic supplementary materialThe online version of this article (10.1186/s13321-018-0287-6) contains supplementary material, which is available to authorized users.

DeepScaffold: A Comprehensive Tool for Scaffold-Based De Novo Drug Discovery Using Deep Learning

“…A scaffold-based molecule generative model; c. A filter based on side-chain properties workflow largely follows that used in our previous study. 18 First of all, the structural information of molecules represented as canonical SMILES are first extracted from ChEMBL dataset. The molecules are then standardized using RDKit, which involves the removal of salt and isotopes, as well as charge neutralization.…”

“…The generation process of side-chain follows largely from our previous work. 18 The major difference is that for scaffold-based generation, the process starts from the graph of the given scaffold, instead of an empty graph (see Figure 7). The model builds the molecule in a step-by-step fashion.…”

“…We use the likelihood-based loss function for model optimization. Importance sampling is used to improve the performance of the model (following 18 ). The value for the hyperparameter k (the number of samples generated from importance sampling) and α (parameter controlling the degree of uncertainty in route sampling) is set as 5 and 0.5 respectively.…”

Inverse‐QSPR for de novo Design: A Review

“…Li et al [157] implemented sequential graph generators called MolMP and MolRNN. To generate molecules those algorithms apply iterative graph transformations.…”

“…Graphs Convolutional Policy Networks [158] are sequential graph generators based on GANs. Molecules are constructed iteratively by choosing the best action to be performed as in the work by Li et al [157] Moreover, subgraphs -fragments -are also considered as new elements for eventual addition. Adding connectivity inside the current graph (i. e., to connect a new subgraph or add an edge between two atoms inside the existing structure) was also possible.…”