Learning to Make Generalizable and Diverse Predictions for Retrosynthesis

TL;DR

This paper introduces a Transformer-based model with novel pre-training and latent variable techniques to improve the accuracy and diversity of retrosynthetic reaction predictions, demonstrating significant gains on the USPTO-50k benchmark.

Contribution

It presents a new model architecture with pre-training strategies and diversity-promoting mechanisms specifically designed for retrosynthesis prediction.

Findings

Significantly outperforms baseline models on USPTO-50k dataset.

Produces more diverse reaction predictions.

Enhances generalizability of retrosynthesis models.

Abstract

We propose a new model for making generalizable and diverse retrosynthetic reaction predictions. Given a target compound, the task is to predict the likely chemical reactants to produce the target. This generative task can be framed as a sequence-to-sequence problem by using the SMILES representations of the molecules. Building on top of the popular Transformer architecture, we propose two novel pre-training methods that construct relevant auxiliary tasks (plausible reactions) for our problem. Furthermore, we incorporate a discrete latent variable model into the architecture to encourage the model to produce a diverse set of alternative predictions. On the 50k subset of reaction examples from the United States patent literature (USPTO-50k) benchmark dataset, our model greatly improves performance over the baseline, while also generating predictions that are more diverse.