We present here the use of QM/MM LIE (linear interaction energy) based binding free energy calculations that greatly improve the precision and accuracy of predicting experimental binding affinities, in comparison to most current binding free energy methodologies, while maintaining reasonable computational times. Calculations are done for four sets of ligand protein complexes, chosen on the basis of diversity of protein types and availability of experimental data, totaling 140 ligands binding to therapeutic protein targets BACE1, TYK2, HSP90, and PERK. This method allows calculations for a diverse set of ligands and multiple protein targets without the need for parametrization or extra calculations. The accuracy achieved with this method can be used to guide small molecule computational drug design programs.