Sampling Strategies in Free Energy Calculation

To understand many important chemical processes, it is often necessary to study their underlying free energy behavior. This is the case, for instance, in studies of protein-ligand binding, solvation, partitioning across the cell membrane, configuration stability and folding. In particular, computer-aided rational drug design relies crucially on accurate prediction of binding free energy between candidate compounds and their target. While its statistical mechanics foundation has been well established, free energy calculation in practical applications is still a very computationally challenging problem. A crucial step in free energy calculation is to sample the vast conformation space of the biomolecule of interest, by using simulation methods such as molecular dynamics and Monte Carlo. The accuracy of free energy calculation depends on how exhaustively the conformation space is sampled. Therefore, given a fixed amount of computer resources, we are usually faced with the issue of how to set up the simulation to achieve the best possible accuracy. In this talk, Nguyen will discuss recent simulation results and analysis of the statistical variance of estimated free energy, which might give a hint on how to go about setting up simulations for an accurate estimation of free energy.