Learning Multimodal Graph-to-Graph Translation for Molecular Optimization

TL;DR

This paper introduces a novel graph-to-graph translation model for molecular optimization, leveraging a junction tree encoder-decoder and adversarial training to generate diverse, property-improved molecules, outperforming existing methods.

Contribution

It proposes a new multimodal graph translation framework with a junction tree encoder-decoder and adversarial training for diverse molecular optimization.

Findings

Outperforms previous state-of-the-art models on multiple tasks

Successfully generates diverse molecules with improved properties

Effectively models multiple translation outputs for each input molecule

Abstract

We view molecular optimization as a graph-to-graph translation problem. The goal is to learn to map from one molecular graph to another with better properties based on an available corpus of paired molecules. Since molecules can be optimized in different ways, there are multiple viable translations for each input graph. A key challenge is therefore to model diverse translation outputs. Our primary contributions include a junction tree encoder-decoder for learning diverse graph translations along with a novel adversarial training method for aligning distributions of molecules. Diverse output distributions in our model are explicitly realized by low-dimensional latent vectors that modulate the translation process. We evaluate our model on multiple molecular optimization tasks and show that our model outperforms previous state-of-the-art baselines.