Abstract
Modeling of transition state by molecular dynamic method often requires modification of the force field parameters to describe energy profile accurately. In this work, we avoided the modification by modeling a series of mutants at binding-related site. In predicting the catalytic efficiency (k cat /K m ) of the mutants of mandelate racemase (MR), the prediction performance of three energy subsets was investigated. It was indicated that the interaction-energy subset exhibited better prediction performance than whole-system subset and binding-site subset in both quantity and trend. When prediction error (PE) criterion was equal to 5%, 10 out of 12 samples were predicted correctly within interaction-energy subset, which demonstrated a great application potential of this method in prediction of enzyme catalytic efficiency and enzyme rational design.