RetroGFN: Diverse and Feasible Retrosynthesis using GFlowNets

TL;DR

RetroGFN is a novel model for single-step retrosynthesis that explores beyond limited datasets to generate diverse, feasible reactions, improving round-trip accuracy and expanding the understanding of reaction feasibility.

Contribution

The paper introduces RetroGFN, a GFlowNet-based model that enhances reaction diversity and feasibility assessment in retrosynthesis, addressing dataset limitations.

Findings

RetroGFN achieves competitive top-k accuracy.

RetroGFN outperforms existing methods on round-trip accuracy.

Empirical evidence supports using round-trip accuracy as a feasibility metric.

Abstract

Single-step retrosynthesis aims to predict a set of reactions that lead to the creation of a target molecule, which is a crucial task in molecular discovery. Although a target molecule can often be synthesized with multiple different reactions, it is not clear how to verify the feasibility of a reaction, because the available datasets cover only a tiny fraction of the possible solutions. Consequently, the existing models are not encouraged to explore the space of possible reactions sufficiently. In this paper, we propose a novel single-step retrosynthesis model, RetroGFN, that can explore outside the limited dataset and return a diverse set of feasible reactions by leveraging a feasibility proxy model during the training. We show that RetroGFN achieves competitive results on standard top-k accuracy while outperforming existing methods on round-trip accuracy. Moreover, we provide empirical arguments in favor of using round-trip accuracy, which expands the notion of feasibility with respect to the standard top-k accuracy metric.