Highly accurate enzyme turnover number prediction and enzyme engineering with PreKcat

<title>Abstract</title> Prediction of enzyme turnover number (<italic>k</italic>_cat) is essential for designing and optimizing enzymes for various biotechnological and industrial applications, but the limited performance of current prediction tools on diverse tasks hinders their practical applications. Here, we introduce PreKcat, a method based on pretrained language models for highly accurate <italic>k</italic>_cat prediction from protein sequences and substrate structures. A two-layer framework derived from PreKcat was proposed to allow robust <italic>k</italic>_cat prediction in considering environmental factors, including pH and temperature. In addition, four representative re-weighting methods were systematically explored to successfully reduce the prediction error in high <italic>k</italic>_cat values prediction tasks. Furthermore, PreKcat showed strong generalizability in two extended tasks, <italic>K</italic>_m and <italic>k</italic>_cat / <italic>K</italic>_mprediction. Using PreKcat, we achieve significant increases <italic>k</italic>_cat and <italic>k</italic>_cat / <italic>K</italic>_m (approximately 7-fold and 3.5-fold, respectively) on one enzyme, the highest reported values to date. PreKcat is a valuable tool for deciphering the mechanisms of enzyme kinetics and enables novel insights into enzyme engineering and their industrial applications.