UC San Diego

Binding free energy determines binding affinity and therefore is central to the design of ligands to bind specific proteins with high affinity. While ever-growing experimental technologies can yield reliable binding affinities, investment of experiments is usually costly. Free energy methods with computer modeling are powerful tools to predict binding affinity and guide directions for experimental input. However, computational methods still are not accurate enough to rely on their own, and sometimes can be computationally demanding. In this thesis, I focused on how to calculate the binding affinity with a fast and accurate approach. I started by evaluating the performance of fast Generalized Born (GB) implicit solvent models in terms of binding free energy calculations with host-guest systems that capture most of the intramolecular and intermolecular interactions of protein-ligand binding complexes. Then, the results from these calculations motivated the second study to explore the possibility of improving the accuracy of such calculations by small adjustments in GB parameters. As a result, this work guides directions for fast and accurate free energy calculations and potential optimization of GB models.