Estimating the inhibitory potency of ligands and catalytic activity of enzyme mutants remains challenging as commonly employed empirical scoring functions still lack sufficient performance and routine in silico screening of mutant libraries with quantum chemical approaches is too computationally extensive. We propose a ligand scoring model MED [1] based on first principles and accounting for two long-range interaction energy components: multipole electrostatic and approximate dispersion terms. By merging computational efficiency with the lack of arbitrary parameterization, it functions as a robust tool for predicting the relative binding energy within enzyme binding and/or active sites. Herein, the predictive capabilities of the MED model will be compared with those of empirical scoring methods. The recent developments in MED-guided determination of mutant catalytic activity will also be presented, along with emerging principles to be followed in computational de novo enzyme design.

Acknowledgements: Support from NCN PRELUDIUM 2016/21/N/ST4/00516 grant is acknowledged. Calculations were performed at the Wroclaw Centre for Networking and Supercomputing.

[1] W. Jedwabny, E. Dyguda-Kazimierowicz, K. Pernal, K. Szalewicz, K. Patkowski, J. Phys. Chem. A 125, 1787 (2021)