Proteins are complex biomolecules that perform a variety of crucial functions within living organisms. Designing and generating novel proteins can pave the way for many future synthetic biology applications, including drug discovery. However, it remains a challenging computational task due to the large modeling space of protein structures. In this study, we propose a latent diffusion model that can reduce the complexity of protein modeling while flexibly capturing the distribution of natural protein structures in a condensed latent space. Specifically, we propose an equivariant protein autoencoder that embeds proteins into a latent space and then uses an equivariant diffusion model to learn the distribution of the latent protein representations. Experimental results demonstrate that our method can effectively generate novel protein backbone structures with high designability and efficiency. The code will be made publicly available at https://github.com/divelab/AIRS/tree/main/OpenProt/LatentDiff
@article{arxiv.2305.04120,
title = {A Latent Diffusion Model for Protein Structure Generation},
author = {Cong Fu and Keqiang Yan and Limei Wang and Wing Yee Au and Michael McThrow and Tao Komikado and Koji Maruhashi and Kanji Uchino and Xiaoning Qian and Shuiwang Ji},
journal= {arXiv preprint arXiv:2305.04120},
year = {2023}
}
Comments
Accepted by the Second Learning on Graphs Conference (LoG 2023)