Quantifying the Role of Water in Ligand-Protein Binding Processes

The aim of this thesis is to quantify the contributions of water thermodynamics to the binding free energy in protein-ligand complexes. Various computational tools were directly applied, implemented, benchmarked and discussed.
An own implementation of the IFST formulation was developed to facilitate easy integration in workflows that are based on Schrödinger software. By applying the tool to a well-defined test set of congeneric ligand pairs, the potential of IFST for quantitative predictions in lead-optimization was assessed.
Furthermore, FEP calculations were applied to an extended test set to validate if these simulations can accurately account for solvent displacement in ligand modifications.
As a fast tool that has applications in virtual screening problems, we finally developed and validated a new scoring function that incorporates terms for protein and ligand desolvation.
This resulted in total in three distinct studies, that all elucidated different aspects of water thermodynamics in CADD. These three studies are presented in the next section. In the conclusion, the results and implications of these studies are discussed jointly, as well with possible future developments.
An additional study was focused on virtual screening and toxicity prediction at the androgen receptor, where distinguishing agonists and antagonists poses difficulties. We proposed and validated an approach based on MD simulations and ensemble docking to improve predictions of androgen agonists and antagonists.