CHA2: CHemistry Aware Convex Hull Autoencoder Towards Inverse Molecular Design

TL;DR

This paper introduces CHA2, a chemistry-aware autoencoder that uses a convex hull in latent space to efficiently generate novel molecules with high drug-likeness scores, addressing the challenge of inverse molecular design.

Contribution

The paper presents a novel convex hull approach in the latent space of an autoencoder for inverse molecular design, improving the discovery of high-QED molecules.

Findings

Effective identification of high-QED molecules using the convex hull method

Demonstrated success on QM9 dataset with SELFIES representation

Enhanced exploration of the molecular design space

Abstract

Optimizing molecular design and discovering novel chemical structures to meet certain objectives, such as quantitative estimates of the drug-likeness score (QEDs), is NP-hard due to the vast combinatorial design space of discrete molecular structures, which makes it near impossible to explore the entire search space comprehensively to exploit de novo structures with properties of interest. To address this challenge, reducing the intractable search space into a lower-dimensional latent volume helps examine molecular candidates more feasibly via inverse design. Autoencoders are suitable deep learning techniques, equipped with an encoder that reduces the discrete molecular structure into a latent space and a decoder that inverts the search space back to the molecular design. The continuous property of the latent space, which characterizes the discrete chemical structures, provides a flexible representation for inverse design in order to discover novel molecules. However, exploring this latent space requires certain insights to generate new structures. We propose using a convex hall surrounding the top molecules in terms of high QEDs to ensnare a tight subspace in the latent representation as an efficient way to reveal novel molecules with high QEDs. We demonstrate the effectiveness of our suggested method by using the QM9 as a training dataset along with the Self- Referencing Embedded Strings (SELFIES) representation to calibrate the autoencoder in order to carry out the Inverse molecular design that leads to unfold novel chemical structure.