3D

Diffusion

• HierDiff: HierDiff is a hierarchical diffusion model for 3D molecule generation, presenting a coarse-to-fine approach that aims to efficiently and effectively generate drug-like molecules in 3D space.
  

  

• JODO: JODO focuses on learning joint 2D and 3D diffusion models for complete molecule generation, representing molecules as both 3D point clouds and 2D bonding graphs to enhance molecular design.
  

  

• MDM: Molecular Diffusion Model for 3D Molecule Generation
  

  

• MolCode: MolCode introduces a roto-translation equivariant generative framework for co-designing molecular 2D graph structures and 3D geometries, aiming to efficiently learn the structure-property relationship for molecule generation.