Scalable Multimer Structure Prediction using Diffusion Models

Published: 28 Oct 2023, Last Modified: 04 Nov 2023NeurIPS2023-AI4Science PosterEveryoneRevisionsBibTeX
Keywords: diffusion generative models, protein complex structure prediction, generative modeling
TL;DR: We propose a new diffusion generative model for faster and more scalable protein complex modeling with accuracy comparable to existing methods.
Abstract: Accurate protein complex structure modeling is a necessary step in understanding the behavior of biological pathways and cellular systems. While some works have attempted to address this challenge, there is still a need for scaling existing methods to larger protein complexes. To address this need, we propose a novel diffusion generative model (DGM) that predicts large multimeric protein structures by learning to rigidly dock its chains together. Additionally, we construct a new dataset specifically for large protein complexes used to train and evaluate our DGM. We substantially improve prediction runtime and completion rates while maintaining competitive accuracy with current methods.
Submission Track: Original Research
Submission Number: 35