It is increasingly clear that we can answer many problems in molecular simulations much more effectively if we consider simulations over a range of different potential functions. Simulation approaches include using ensembles of simulations, such as in Hamiltonian replica exchange, or single simulations that visit multiple states, such as Wang-Landau or expanded ensemble simulations. Such approaches can dramatically improve sampling by avoiding kinetic barriers that disappear at some values of the potential energy variable we consider, and can improve estimation of thermodynamic properties as a function of a range of potential energy parameter values through the use of reweighting methods.<br/><br/>I explore a number of avenues of work in our group that help us to better understand simulations using the potential energy as a parameter, including the multistate Bennett acceptance ratio for calculating free energies, expectation values, and their uncertainties, as well as the use of Gibbs sampling to accelerate dynamics in potential energy variables. I discuss a number of applications that our group has investigated that have benefitted from the power of treating the potential energy as a variable parameter, including accelerating simulations of the free energy of binding, exploring the full binding ensemble of weak ligands, and multidimensional parameterization of force fields.

Potential function as a variable: Advances in using simulations of multiple states to solve hard biomolecular problems

249th National Meeting (2015)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.