Conditional molecular design with deep generative models

TL;DR

This paper introduces a semi-supervised variational autoencoder model for conditional molecular design, enabling efficient generation of novel molecules with desired properties by leveraging both labeled and unlabeled data.

Contribution

It presents a novel semi-supervised generative model that combines property prediction and molecule generation, improving efficiency and accuracy in chemical space exploration.

Findings

Enhanced property prediction accuracy using unlabeled data

Efficient generation of molecules with specified properties

Validated on drug-like molecules with promising results

Abstract

Although machine learning has been successfully used to propose novel molecules that satisfy desired properties, it is still challenging to explore a large chemical space efficiently. In this paper, we present a conditional molecular design method that facilitates generating new molecules with desired properties. The proposed model, which simultaneously performs both property prediction and molecule generation, is built as a semi-supervised variational autoencoder trained on a set of existing molecules with only a partial annotation. We generate new molecules with desired properties by sampling from the generative distribution estimated by the model. We demonstrate the effectiveness of the proposed model by evaluating it on drug-like molecules. The model improves the performance of property prediction by exploiting unlabeled molecules, and efficiently generates novel molecules fulfilling various target conditions.