Computational Approaches to the Chemical Equilibrium Constant in Protein‐ligand Binding

The physiological role played by protein‐ligand recognition has motivated the development of several computational approaches to the ligand binding affinity. Some of them, termed rigorous, have a strong theoretical foundation but involve too much computation to be generally useful. Some others alleviate the computational burden by introducing strong approximations and/or empirical calibrations, which also limit their general use. Most importantly, there is no straightforward correlation between the predictive power and the level of approximation introduced. Here, we present a general framework for the quantitative interpretation of protein‐ligand binding based on statistical mechanics. Within this framework, we re‐derive self‐consistently the fundamental equations of some popular approaches to the binding constant and pinpoint the inherent approximations. Our analysis represents a first step towards the development of variants with optimum accuracy/efficiency ratio for each stage of the drug discovery pipeline.