Trustworthy prediction of enzyme commission numbers using a hierarchical interpretable transformer.

Abstract

Accurate and trustworthy prediction of Enzyme Commission (EC) numbers is critical for understanding enzyme functions and their roles in biological processes. Despite the success of recently proposed deep learning-based models, there remain limitations, such as low performance in underrepresented EC numbers, lack of learning strategy with incomplete annotations, and limited interpretability. To address these challenges, we propose a hierarchical interpretable transformer model, HIT-EC, for trustworthy EC number prediction. HIT-EC employs a four-level transformer architecture that aligns with the hierarchical structure of EC numbers, and leverages both local and global dependencies within protein sequences for this multi-label classification task. We also propose a learning strategy to handle samples associated with incomplete EC numbers. HIT-EC, as an evidential deep learning model, produces trustworthy predictions by providing domain-specific evidence through a biologically meaningful interpretation scheme. The predictive performance of HIT-EC is assessed by multiple experiments: a cross-validation with a large dataset, a validation with external data, and a species-based performance evaluation. HIT-EC shows statistically significant improvement in predictive performance when compared to the current state-of-the-art benchmark models. HIT-EC's robust interpretability is further validated by identifying well-known conserved motifs and functional regions. HIT-EC is a robust, interpretable, and reliable solution for EC number prediction, with significant implications for enzymology, drug discovery, and metabolic engineering.