Fragment-based Sequential Translation for Molecular Optimization

TL;DR

This paper introduces FaST, a fragment-based reinforcement learning method for molecular optimization that outperforms existing atom-based approaches by generating novel molecules through learned substructure editing.

Contribution

It proposes a novel fragment-based editing paradigm using a VAE for encoding molecular fragments and a RL policy for sequential molecule translation, advancing molecular design techniques.

Findings

FaST significantly outperforms state-of-the-art methods on benchmark tasks.

The learned fragment vocabulary enables more meaningful molecule editing.

The approach effectively explores complex chemical property spaces.

Abstract

Searching for novel molecular compounds with desired properties is an important problem in drug discovery. Many existing frameworks generate molecules one atom at a time. We instead propose a flexible editing paradigm that generates molecules using learned molecular fragments--meaningful substructures of molecules. To do so, we train a variational autoencoder (VAE) to encode molecular fragments in a coherent latent space, which we then utilize as a vocabulary for editing molecules to explore the complex chemical property space. Equipped with the learned fragment vocabulary, we propose Fragment-based Sequential Translation (FaST), which learns a reinforcement learning (RL) policy to iteratively translate model-discovered molecules into increasingly novel molecules while satisfying desired properties. Empirical evaluation shows that FaST significantly improves over state-of-the-art methods on benchmark single/multi-objective molecular optimization tasks.