In Silico Methods for Ranking Ligand–Protein Interactions and Predicting Binding Affinities: Which Method is Right for You?
Stuart Lang, Natércia F. Braz, Martin J. Slater, Nathan J. Kidley · Journal of Medicinal Chemistry · 2025
M easuring a ligand's ability to bind to a protein is a central pillar in all drug discovery programs, both in terms of maximizing affinity for the primary target while also increasing selectivity by minimizing binding to undesired proteins.1 Ligand Optimization involves multiple iterations of the design-make-test-analysis process, with hundreds, maybe thousands, of compounds often requiring synthesis and biological evaluation in multiple assays to bring a potential drug to the clinic.While requiring experimental data to build and train models, computational methods have significantly improved the efficiency of prioritizing ligands at the design phase and in triaging virtual screening 2 outputs, meaning fewer compounds are required for experimental biological testing to progress a project.3 As computational methods evolve, with new tools continually being developed, it is important to understand which method is optimal for each situation, both in terms of accuracy and required computational resources (Table 1). ■ METHODS ANALYZING LIGAND-PROTEIN INTERACTIONSOne of the simplest, and best known, methods for assessment of a ligand's binding to a protein is molecular docking.4 This method is primarily used to predict how a ligand will bind to a protein in 3D.It is fast with a low computational cost associated with it.Docking scores are reported as a Gibbs Free Energy (ΔG), with a higher negative value indicating stronger binding energy of the ligand to the specified protein.Docking algorithms often generate multiple poses for each ligand, with the best scoring pose not always being the correct binding pose.This means that assessment of the binding pose should not be made exclusively on docking score.A useful docking validation protocol, if compounds of known activity are available, is to generate scored docking poses for these ligands in the protein being studied.This allows the docking scores to be compared to the true experimental values, and a judgment made to the validity of these docking scores and/or the proposed binding pose.Carrying out this assessment on known ligands allows optimization of the docking protocol and increases the confidence of this scoring method for prioritizing new ligand designs prior to synthesis.Docking scores are accumulative, meaning that larger ligands making more interactions with the protein will generally score better than smaller ligands.For this reason, docking Ligand Efficiency (LE) metrics, like experimental LE, 5 can be applied to allow molecules of differing size to be compared.Electrostatic Complementarity (EC) 6 is also an inexpensive