Using GNN property predictors as molecule generators

TL;DR

This paper presents a novel method that uses invertible graph neural networks to directly generate molecular structures with specific electronic properties by optimizing input graphs through gradient ascent, without additional training.

Contribution

The authors introduce a property-guided molecule generation technique leveraging GNN invertibility, enabling direct optimization of molecular graphs for desired properties without extra training.

Findings

Achieves target properties with high accuracy and diversity.

Outperforms or matches state-of-the-art generative models.

Generates molecules with verified electronic properties.

Abstract

Graph neural networks (GNNs) have emerged as powerful tools to accurately predict materials and molecular properties in computational discovery pipelines. In this article, we exploit the invertible nature of these neural networks to directly generate molecular structures with desired electronic properties. Starting from a random graph or an existing molecule, we perform a gradient ascent while holding the GNN weights fixed in order to optimize its input, the molecular graph, towards the target property. Valence rules are enforced strictly through a judicious graph construction. The method relies entirely on the property predictor; no additional training is required on molecular structures. We demonstrate the application of this method by generating molecules with specific DFT-verified energy gaps and octanol-water partition coefficients (logP). Our approach hits target properties with rates comparable to or better than state-of-the-art generative models while consistently generating more diverse molecules.