Geometric Latent Diffusion Models for 3D Molecule Generation
Geometric Latent Diffusion Models for 3D Molecule Generation
复制标题
DOI：
10.48550/arxiv.2305.01140
复制
发表时间：
2023-05
期刊：
影响因子：
--
通讯作者：
Minkai Xu;Alexander Powers;R. Dror;Stefano Ermon;J. Leskovec
中科院分区：
文献类型：
--
作者：
Minkai Xu;Alexander Powers;R. Dror;Stefano Ermon;J. Leskovec
生成模型，特别是扩散模型(DM)，在生成特征丰富的几何图形和推进分子设计等基础科学问题方面取得了可喜的成果。受近年来稳定(潜在)扩散模型的巨大成功的启发，我们提出了一种新的、原则性的三维分子生成方法--几何潜在扩散模型(GeoLDM)。GeoLDM是分子几何领域的第一个潜在DM模型，由自动编码器将结构编码成连续的潜在代码和在潜在空间中操作的DM组成。我们的主要创新之处在于，对于三维分子几何模型，我们通过构建具有不变标量和等变张量的点结构的潜在空间来捕捉其关键的旋转-平移等变约束。广泛的实验表明，GeoLDM在多个分子生成基准上可以持续获得更好的性能，大分子的有效百分比提高高达7%。结果还表明，由于潜在的建模，GeoLDM具有更高的可控发电能力。代码位于\url{https://github.com/MinkaiXu/GeoLDM}.
Generative models, especially diffusion models (DMs), have achieved promising results for generating feature-rich geometries and advancing foundational science problems such as molecule design. Inspired by the recent huge success of Stable (latent) Diffusion models, we propose a novel and principled method for 3D molecule generation named Geometric Latent Diffusion Models (GeoLDM). GeoLDM is the first latent DM model for the molecular geometry domain, composed of autoencoders encoding structures into continuous latent codes and DMs operating in the latent space. Our key innovation is that for modeling the 3D molecular geometries, we capture its critical roto-translational equivariance constraints by building a point-structured latent space with both invariant scalars and equivariant tensors. Extensive experiments demonstrate that GeoLDM can consistently achieve better performance on multiple molecule generation benchmarks, with up to 7\% improvement for the valid percentage of large biomolecules. Results also demonstrate GeoLDM's higher capacity for controllable generation thanks to the latent modeling. Code is provided at \url{https://github.com/MinkaiXu/GeoLDM}.