MolHIT: Advancing Molecular-Graph Generation with Hierarchical Discrete Diffusion Models

TL;DR

MolHIT introduces a hierarchical discrete diffusion framework for molecular graph generation, significantly improving validity and property control, and setting new state-of-the-art results in AI-driven drug discovery tasks.

Contribution

It proposes a novel hierarchical discrete diffusion model with chemical priors and decoupled atom encoding, overcoming previous limitations in molecular graph generation.

Findings

Achieves near-perfect validity in molecular graph generation.

Sets new state-of-the-art performance on the MOSES dataset.

Excels in downstream tasks like multi-property guided generation.

Abstract

Molecular generation with diffusion models has emerged as a promising direction for AI-driven drug discovery and materials science. While graph diffusion models have been widely adopted due to the discrete nature of 2D molecular graphs, existing models suffer from low chemical validity and struggle to meet the desired properties compared to 1D modeling. In this work, we introduce MolHIT, a powerful molecular graph generation framework that overcomes long-standing performance limitations in existing methods. MolHIT is based on the Hierarchical Discrete Diffusion Model, which generalizes discrete diffusion to additional categories that encode chemical priors, and decoupled atom encoding that splits the atom types according to their chemical roles. Overall, MolHIT achieves new state-of-the-art performance on the MOSES dataset with near-perfect validity for the first time in graph diffusion, surpassing strong 1D baselines across multiple metrics. We further demonstrate strong performance in downstream tasks, including multi-property guided generation and scaffold extension.