{"title":"Geometry-complete diffusion for 3D molecule generation and optimization","authors":"Alex Morehead, Jianlin Cheng","doi":"10.1038/s42004-024-01233-z","DOIUrl":null,"url":null,"abstract":"Generative deep learning methods have recently been proposed for generating 3D molecules using equivariant graph neural networks (GNNs) within a denoising diffusion framework. However, such methods are unable to learn important geometric properties of 3D molecules, as they adopt molecule-agnostic and non-geometric GNNs as their 3D graph denoising networks, which notably hinders their ability to generate valid large 3D molecules. In this work, we address these gaps by introducing the Geometry-Complete Diffusion Model (GCDM) for 3D molecule generation, which outperforms existing 3D molecular diffusion models by significant margins across conditional and unconditional settings for the QM9 dataset and the larger GEOM-Drugs dataset, respectively. Importantly, we demonstrate that GCDM’s generative denoising process enables the model to generate a significant proportion of valid and energetically-stable large molecules at the scale of GEOM-Drugs, whereas previous methods fail to do so with the features they learn. Additionally, we show that extensions of GCDM can not only effectively design 3D molecules for specific protein pockets but can be repurposed to consistently optimize the geometry and chemical composition of existing 3D molecules for molecular stability and property specificity, demonstrating new versatility of molecular diffusion models. Code and data are freely available on GitHub . Geometric deep learning methods have the advantage of being expressive, while denoising diffusion probabilistic models have great generative power. Here, the authors introduce a geometry-complete diffusion model for effective 3D molecule generation for specific protein pockets that can also consistently optimize the geometry and chemical composition of existing 3D molecules for molecular stability and property specificity","PeriodicalId":10529,"journal":{"name":"Communications Chemistry","volume":null,"pages":null},"PeriodicalIF":5.9000,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11222514/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Communications Chemistry","FirstCategoryId":"92","ListUrlMain":"https://www.nature.com/articles/s42004-024-01233-z","RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CHEMISTRY, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Generative deep learning methods have recently been proposed for generating 3D molecules using equivariant graph neural networks (GNNs) within a denoising diffusion framework. However, such methods are unable to learn important geometric properties of 3D molecules, as they adopt molecule-agnostic and non-geometric GNNs as their 3D graph denoising networks, which notably hinders their ability to generate valid large 3D molecules. In this work, we address these gaps by introducing the Geometry-Complete Diffusion Model (GCDM) for 3D molecule generation, which outperforms existing 3D molecular diffusion models by significant margins across conditional and unconditional settings for the QM9 dataset and the larger GEOM-Drugs dataset, respectively. Importantly, we demonstrate that GCDM’s generative denoising process enables the model to generate a significant proportion of valid and energetically-stable large molecules at the scale of GEOM-Drugs, whereas previous methods fail to do so with the features they learn. Additionally, we show that extensions of GCDM can not only effectively design 3D molecules for specific protein pockets but can be repurposed to consistently optimize the geometry and chemical composition of existing 3D molecules for molecular stability and property specificity, demonstrating new versatility of molecular diffusion models. Code and data are freely available on GitHub . Geometric deep learning methods have the advantage of being expressive, while denoising diffusion probabilistic models have great generative power. Here, the authors introduce a geometry-complete diffusion model for effective 3D molecule generation for specific protein pockets that can also consistently optimize the geometry and chemical composition of existing 3D molecules for molecular stability and property specificity
期刊介绍:
Communications Chemistry is an open access journal from Nature Research publishing high-quality research, reviews and commentary in all areas of the chemical sciences. Research papers published by the journal represent significant advances bringing new chemical insight to a specialized area of research. We also aim to provide a community forum for issues of importance to all chemists, regardless of sub-discipline.