Probing Limitations of Co-Alchemical Charge Changes in Free-Energy Calculations

Molecular dynamics simulations are nowadays one of the key methods to investigate the (thermo)dynamics of protein-ligand binding at atomic resolution. The calculation of binding free energies of charged species is an encountered problem in molecular dynamic simulations. This is due to the approximation of the long-range electrostatic interaction. Here, we explore the discrepancies and biases of different approaches and whether and under which circumstances robust and reliable free-energy differences can be obtained using alchemical methods in combination with the lattice sum electrostatics treatment. Testing various setups and well-established approaches shows that the obtained free energies strongly depend on the initial setup choices. Different unconventional schemes, for example, placing more copies of perturbed species in a simulation box, were tested and showed partial success. Although they still suffer from pitfalls, they present a promising alternative approach in addressing the challenges related to non-neutral perturbations.