POTENTIALS OF MEAN FORCE FOR BIOMOLECULAR SIMULATIONS - THEORY AND TEST ON ALANINE DIPEPTIDE

We describe a technique for generating potentials of mean force (PMF) between solutes in an aqueous solution. We first generate solute-solve nt correlation functions (CF) using Monte Carlo (MC) simulations in wh ich we place a single atom solute in a periodic boundary box containin g a few hundred water molecules. We then make use of the Kirkwood supe rposition approximation, where the 3-body correlation function is appr oximated as the product of 2-body CFs, to describe the mean water dens ity around two solutes. Computing the force generated on the solutes b y this average water density allows us to compute potentials of mean f orce between the two solutes. For charged solutes an additional approx imation involving dielectric screening is made, by setting the dielect ric constant of water to epsilon=80. These potentials account, in an a pproximate manner, for the average effect of water on the atoms. Follo wing the work of Pettitt and Karplus [Chem. Phys. Lett. 121, 194 (1985 )], we approximate the n-body potential of mean force as a sum of the pairwise potentials of mean force. This allows us to run simulations o f biomolecules without introducing explicit water, hence gaining sever al orders of magnitude in efficiency with respect to standard molecula r dynamics techniques. We demonstrate the validity of this technique b y first comparing the PMFs for methane-methane and sodium-chloride gen erated with this procedure, with those calculated with a standard Mont e Carlo simulation with explicit water. We then compare the results of the free energy profiles between the equilibria of alanine dipeptide generated by the two methods. (C) 1996 American Institute of Physics.