A Benchmark Test Set for Free Energy Methods and Its Use to Test Free Energy Estimators and Dependence On Molecular Simulation Parameters

We recently developed a benchmark set for testing methods to calculate free energies of molecular transformations in solution, based aroundmolecular changes common to many molecular design problems such as atomic insertion and/or deletion and changing atomic partial charges. This benchmark set is provided in a number of different file formats with the hope of becoming a useful and general benchmark.  We describe the use of this benchmark set to compare the statistical efficiency of free energy estimates, demonstrating that MBAR has superior statistical properties to other free energy estimation methods.  We also use the benchmark set to analyze the effect ofchanges in simulation parameters such as PME parameters and dispersion cutoffs, as well as differences between molecular simulation programs.