MiDi: Mixed Graph and 3D Denoising Diffusion for Molecule Generation

Clement Vignac, Nagham Osman, Laura Toni, Pascal Frossard
2/17/2023
cs.LG

No official code found. Best implementation: cvignac/MiDi (116 stars, MIT, last commit 2024-09-18).

Abstract

This work introduces MiDi, a novel diffusion model for jointly generating molecular graphs and their corresponding 3D arrangement of atoms. Unlike existing methods that rely on predefined rules to determine molecular bonds based on the 3D conformation, MiDi offers an end-to-end differentiable approach that streamlines the molecule generation process. Our experimental results demonstrate the effectiveness of this approach. On the challenging GEOM-DRUGS dataset, MiDi generates 92% of stable molecules, against 6% for the previous EDM model that uses interatomic distances for bond prediction, and 40% using EDM followed by an algorithm that directly optimize bond orders for validity. Our code is available at github.com/cvignac/MiDi.

View on arXivView PDF

Code Implementations(1)

cvignac/MiDiCommunity100%
MIT

MiDi: Mixed Graph and 3D Denoising Diffusion for Molecule Generation

11616PythonFeb 17, 20232 years agoMIT

Similar Papers

Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation

87% match1 repo102024

Learning Joint 2D & 3D Diffusion Models for Complete Molecule Generation

87% match1 repo552023

MUDiff: Unified Diffusion for Complete Molecule Generation

87% match2023

3M-Diffusion: Latent Multi-Modal Diffusion for Language-Guided Molecular Structure Generation

86% match1 repo282024

Geometry-Complete Diffusion for 3D Molecule Generation and Optimization

86% match2 repos2422023

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

84% match2026

Cite this paper

@article{vignac2023midi,
  title  = {MiDi: Mixed Graph and 3D Denoising Diffusion for Molecule Generation},
  author = {Clement Vignac and Nagham Osman and Laura Toni and Pascal Frossard},
  year   = {2023},
  eprint = {2302.09048},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2302.09048v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2302.09048.svg)](https://codeofpaper.com/paper/2302.09048)

Discussion