Accurate ranking of binding affinities is crucial in the lead optimization phase of pharmaceutical research in order to develop potent, effective drug candidates. Both academic groups and the pharmaceutical industry have invested a great deal of effort to meet this challenge. Several approaches have been developed, ranging from rapid QSAR-based scoring functions to computationally intensive free energy perturbation (FEP) calculations. But none have fully met the needs of researcher and developers. QSAR-type approaches, though rapid, involve many approximations and produce large errors in binding energy predictions. FEP approaches are more accurate, but cannot be used when ligand structures vary significantly. They also incur substantial CPU costs.
Linear interaction approximation (LIA) is a way of combining molecular mechanics calculations with experimental data to build a model scoring function for the evaluation of ligand-protein binding free energies. LIA methods strike a perfect balance between accuracy and computational cost.