Advancing Protein Function Annotation with Equivariant Graph Networks

Protein function research helps in understanding complex biological processes within cells. However, the intricate nature of protein structures and functions, along with the rapid growth of protein sequence data, presents a pressing challenge to develop efficient computational methods for protein annotation. In this study, we propose ENGINE, a multi-channel deep learning framework designed for robust protein function annotation. ENGINE integrates an equivariant graph convolutional network model to capture geometric features from protein 3D structurals, leverages ESM-C to encode evolutionary and sequence-derived information, and combines an innovative 3D sequence representation that unifies spatial and sequential signals. We demonstrate that ENGINE consistently surpasses current state-of-the-art methods across diverse protein function prediction benchmarks, demonstrating robust generalization and high predictive accuracy. Beyond performance, ENGINE provides interpretable insights into key sequence features and substructures, enabling the identification of functionally critical residues within proteins. This facilitates a deeper mechanistic understanding of protein function annotation outcomes and supports hypothesis generation for downstream biological studies. By offering reliable predictions with biological interpretability, ENGINE contributes to advancing research into cellular processes and disease mechanisms.