OctMole generates 3D molecules precisely from coarse to fine scales
OctMole: Multiscale 3D Molecular Generation with Hierarchical Spatial Flows
Machine Learning
Summary
Making 3D models of molecules is tricky because their overall shape and tiny details like bonds need different levels of focus. The authors created OctMole, a method that builds molecules step-by-step from big shapes down to small atomic details using a special tree-like structure. This approach is faster and more accurate than older methods that look at all atoms equally. OctMole also allows editing parts or blending shapes without extra work.
What this means in practice
- •For drug discovery teams: Automatically generate realistic 3D drug-like molecules with fine atomic precision, improving molecule design workflows.$Commercial implications: This paper enables tools that produce accurate 3D drug candidates efficiently for pharmaceutical companies seeking faster compound design.
- •For biomolecular modelers: Edit and shape 3D molecular regions interactively without retraining, supporting tasks like linker design and structure interpolation.
Authors
Yu-Tong Jin, Yi-Yang Tian, Qiwei Ye, Peng-Shuai Wang
Abstract
Generating accurate 3D molecules requires jointly determining a variable number of atoms, their identities, and geometry across spatial scales: global shape emerges coarsely, whereas chemical bonds demand fine local precision. Existing atom-based diffusion and flow models operate largely at a single scale, while dense spatial fields become costly at high resolution. We introduce OctMole, a multiscale octree flow framework for 3D molecular generation. A hierarchical occupancy flow progressively resolves sparse molecular support from coarse to fine, and a joint atom flow assigns atom types and continuous in-cell offsets to occupied leaves. This decomposition combines efficient global structure with subcell geometric precision, while cell corruption improves robustness to generated support. On QM9 and GEOM-DRUG, OctMole achieves strong chemical, distributional, and geometric quality, with particularly clear gains over Uni-3DAR on GEOM-DRUG and substantial efficiency gains over dense 3D generation. As an additional capability, intermediate occupancy states and selected regions provide spatial conditioning without task-specific retraining, supporting multiscale shape conditioning, regional editing, linker generation, and shape interpolation. Code and pretrained checkpoints are available at https://github.com/octree-nn/octmole and https://huggingface.co/Disembo/octmole, respectively.