A systematic review of EC number prediction algorithms based on reaction chemical structures.

A systematic review compares EC number prediction algorithms that infer enzyme reaction classes from reaction chemical structures, evaluating performance across databases and enzyme classes.

• Why it matters: Incomplete EC annotation in biochemical reaction databases limits accurate mapping of metabolism, hindering functional interpretation in areas like biotechnology and drug development.
• What they did: The review surveys reaction similarity-based and machine/deep learning methods, benchmarking their EC-number prediction performance across multiple databases and testing robustness across enzyme classes.
• The result: The authors also apply these approaches to drug metabolism reactions, suggesting they can improve functional annotation of uncharacterized metabolic pathways.

AI summary of the abstract · may contain errors · not medical advice · read the paper.