Mask prior-guided denoising diffusion improves inverse protein folding

TL;DR

This paper introduces MapDiff, a novel diffusion-based framework that improves inverse protein folding by effectively predicting uncertain regions and capturing structural details, leading to superior sequence design performance.

Contribution

MapDiff is the first discrete diffusion model for inverse protein folding that integrates structural information and residue interactions through a graph-based denoising network with mask-prior pre-training.

Findings

Outperforms state-of-the-art methods on four benchmarks.

Generated sequences closely match native protein characteristics.

Effectively predicts high-uncertainty disordered regions.

Abstract

Inverse protein folding generates valid amino acid sequences that can fold into a desired protein structure, with recent deep-learning advances showing strong potential and competitive performance. However, challenges remain, such as predicting elements with high structural uncertainty, including disordered regions. To tackle such low-confidence residue prediction, we propose a Mask-prior-guided denoising Diffusion (MapDiff) framework that accurately captures both structural information and residue interactions for inverse protein folding. MapDiff is a discrete diffusion probabilistic model that iteratively generates amino acid sequences with reduced noise, conditioned on a given protein backbone. To incorporate structural information and residue interactions, we develop a graph-based denoising network with a mask-prior pre-training strategy. Moreover, in the generative process, we combine the denoising diffusion implicit model with Monte-Carlo dropout to reduce uncertainty. Evaluation on four challenging sequence design benchmarks shows that MapDiff substantially outperforms state-of-the-art methods. Furthermore, the in silico sequences generated by MapDiff closely resemble the physico-chemical and structural characteristics of native proteins across different protein families and architectures.