Annotation-guided Protein Design with Multi-Level Domain Alignment

TL;DR

This paper introduces PAAG, a multi-modality framework that uses textual protein domain annotations to guide the controllable design of proteins with specific functions, achieving higher success rates in protein generation tasks.

Contribution

The paper presents a novel protein design method that explicitly incorporates textual domain annotations for controllable and flexible protein generation, outperforming existing models.

Findings

PAAG significantly improves protein generation success rates.

Explicit annotation alignment enhances controllability in protein design.

Experimental results outperform existing models in key prediction tasks.

Abstract

The core challenge of de novo protein design lies in creating proteins with specific functions or properties, guided by certain conditions. Current models explore to generate protein using structural and evolutionary guidance, which only provide indirect conditions concerning functions and properties. However, textual annotations of proteins, especially the annotations for protein domains, which directly describe the protein's high-level functionalities, properties, and their correlation with target amino acid sequences, remain unexplored in the context of protein design tasks. In this paper, we propose Protein-Annotation Alignment Generation, PAAG, a multi-modality protein design framework that integrates the textual annotations extracted from protein database for controllable generation in sequence space. Specifically, within a multi-level alignment module, PAAG can explicitly generate proteins containing specific domains conditioned on the corresponding domain annotations, and can even design novel proteins with flexible combinations of different kinds of annotations. Our experimental results underscore the superiority of the aligned protein representations from PAAG over 7 prediction tasks. Furthermore, PAAG demonstrates a significant increase in generation success rate (24.7% vs 4.7% in zinc finger, and 54.3% vs 22.0% in the immunoglobulin domain) in comparison to the existing model. We anticipate that PAAG will broaden the horizons of protein design by leveraging the knowledge from between textual annotation and proteins.