Geometry-Complete Diffusion for 3D Molecule Generation-猫眼课题宝

DOI：

10.48550/arxiv.2302.04313

发表时间：

2023

期刊：

ArXiv

影响因子：

通讯作者：

中科院分区：

其他

文献类型：

作者：

Alex Morehead;Jianlin Cheng

去噪扩散概率模型（ddpm）（Ho等人（2020））最近在生成建模领域掀起了一场风暴，在计算机视觉和计算生物学等学科中开创了新的最先进成果，用于各种任务，从文本引导的图像生成(Ramesh等人（2022）；撒哈拉等人（2022年）；Rombach et al.(2022))转向结构导向蛋白设计(Ingraham et al. (2022)；Watson等人（2022）)。沿着后一条研究路线，已经提出了Hoogeboom等人（2022）的方法，用于在DDPM框架内使用等变图神经网络（gnn）生成3D分子。为此，我们提出了GCDM，这是一种几何完全扩散模型，通过利用执行几何完全消息传递的gnn提供的表示学习优势，实现了3D分子扩散生成和优化的最新成果。我们对GCDM的研究结果也为物理归纳偏差如何影响分子ddpm的生成动力学提供了初步的见解。训练新模型或复制我们的结果的源代码、数据和说明可以在https://github.com/BioinfoMachineLearning/Bio-Diffusion上免费获得。

Denoising diffusion probabilistic models (DDPMs) (Ho et al. (2020)) have recently taken the field of generative modeling by storm, pioneering new state-ofthe-art results in disciplines such as computer vision and computational biology for diverse tasks ranging from text-guided image generation (Ramesh et al. (2022); Saharia et al. (2022); Rombach et al. (2022)) to structure-guided protein design (Ingraham et al. (2022); Watson et al. (2022)). Along this latter line of research, methods such as those of Hoogeboom et al. (2022) have been proposed for generating 3D molecules using equivariant graph neural networks (GNNs) within a DDPM framework. Toward this end, we propose GCDM, a geometry-complete diffusion model that achieves new state-of-the-art results for 3D molecule diffusion generation and optimization by leveraging the representation learning strengths offered by GNNs that perform geometry-complete message-passing. Our results with GCDM also offer preliminary insights into how physical inductive biases impact the generative dynamics of molecular DDPMs. The source code, data, and instructions to train new models or reproduce our results are freely available at https://github.com/BioinfoMachineLearning/Bio-Diffusion.

Geometry-Complete Diffusion for 3D Molecule Generation

文献摘要

相似文献