Free Energy Perturbation and Free-Energy Calculations Applied to Drug Design

Free energy perturbation (FEP) is a computational technique used to evaluate ligand-protein binding affinities for computer-aided drug optimization. FEP has been shown to be a valuable tool in both academic and pharmaceutical settings for optimizing drug candidates in a rational, efficient, and cost-effective manner. Recent advancements in algorithms, software tools, hardware capabilities, and machine learning integration have significantly improved the scope, applicability, and reliability of FEP calculations. In this chapter, we review recent developments in force field parameterization, software platforms, and automated workflows that have consolidated FEP as an essential methodology for structure-based drug discovery and have resulted in FEP calculations becoming more accessible to nonspecialists, as well as applicable to a broad range of scenarios. We also describe the utility of the FEP technique in diverse contexts, including validating the binding modes and optimizing allosteric and covalent inhibitors. We illustrate its potential through vignettes and in-depth case studies, demonstrating its integration into machine-learning frameworks for predicting binding energies based on molecular structures. Furthermore, this chapter discusses the remaining challenges in sampling sufficiency and scalability to ultra-large compound libraries as well as emerging solutions through cloud computing and machine learning.