Improving the LIE Method for Binding Free Energy Calculations of Protein-Ligand Complexes

In this work, we introduced an improved linear interaction energy (LIE) method parameterization for computations of protein–ligand binding free energies. The protocol, coined LIE-D, builds on the linear relationship between the empirical coefficient γ in the standard LIE scheme and the D parameter, introduced in our work. The D-parameter encompasses the balance (difference) between electrostatic (polar) and van der Waals (nonpolar) energies in protein–ligand complexes. Leave-one-out cross-validation showed that LIE-D reproduced accurately the absolute binding free energies for our training set of protein–ligand complexes (<|error|> = 0.92 kcal/mol, SDerror = 0.66 kcal/mol, R(2) = 0.90, QLOO(2) = 0.89, and sPRESS(LOO) = 1.28 kcal/mol). We also demonstrated LIE-D robustness by predicting accurately the binding free energies for three different protein–ligand systems outside the training data set, where the electrostatic and van der Waals interaction energies were calculated with different force fields.