PiFold: Toward effective and efficient protein inverse folding

TL;DR

PiFold introduces a novel protein inverse folding method that significantly improves recovery accuracy and inference speed by using a new residue featurizer and PiGNN layers, enabling effective protein design.

Contribution

The paper presents PiFold, a new one-shot protein sequence generator with a novel featurizer and PiGNN layers, achieving high accuracy and speed improvements over existing methods.

Findings

Achieves 51.66% recovery on CATH 4.2

Inference speed is 70 times faster than autoregressive methods

Achieves 58.72% and 60.42% recovery on TS50 and TS500 datasets

Abstract

How can we design protein sequences folding into the desired structures effectively and efficiently? AI methods for structure-based protein design have attracted increasing attention in recent years; however, few methods can simultaneously improve the accuracy and efficiency due to the lack of expressive features and autoregressive sequence decoder. To address these issues, we propose PiFold, which contains a novel residue featurizer and PiGNN layers to generate protein sequences in a one-shot way with improved recovery. Experiments show that PiFold could achieve 51.66\% recovery on CATH 4.2, while the inference speed is 70 times faster than the autoregressive competitors. In addition, PiFold achieves 58.72\% and 60.42\% recovery scores on TS50 and TS500, respectively. We conduct comprehensive ablation studies to reveal the role of different types of protein features and model designs, inspiring further simplification and improvement. The PyTorch code is available at \href{https://github.com/A4Bio/PiFold}{GitHub}.