Coarse-to-Fine: a Hierarchical Diffusion Model for Molecule Generation in 3D

TL;DR

This paper introduces HierDiff, a hierarchical diffusion model that generates 3D molecular structures by progressively refining coarse-grained geometries into detailed atom-level molecules, improving quality over existing methods.

Contribution

It proposes a novel coarse-to-fine hierarchical diffusion approach for 3D molecule generation that preserves local structures without autoregressive modeling.

Findings

HierDiff outperforms existing methods in molecule quality.

The model effectively captures local structures like rings.

It demonstrates robustness on large molecules.

Abstract

Generating desirable molecular structures in 3D is a fundamental problem for drug discovery. Despite the considerable progress we have achieved, existing methods usually generate molecules in atom resolution and ignore intrinsic local structures such as rings, which leads to poor quality in generated structures, especially when generating large molecules. Fragment-based molecule generation is a promising strategy, however, it is nontrivial to be adapted for 3D non-autoregressive generations because of the combinational optimization problems. In this paper, we utilize a coarse-to-fine strategy to tackle this problem, in which a Hierarchical Diffusion-based model (i.e.~HierDiff) is proposed to preserve the validity of local segments without relying on autoregressive modeling. Specifically, HierDiff first generates coarse-grained molecule geometries via an equivariant diffusion process, where each coarse-grained node reflects a fragment in a molecule. Then the coarse-grained nodes are decoded into fine-grained fragments by a message-passing process and a newly designed iterative refined sampling module. Lastly, the fine-grained fragments are then assembled to derive a complete atomic molecular structure. Extensive experiments demonstrate that HierDiff consistently improves the quality of molecule generation over existing methods