THE OVERLOOKED BOND-STRETCHING CONTRIBUTION IN FREE-ENERGY PERTURBATION CALCULATIONS

We examine the validity of an assumption commonly made when carrying out free energy perturbation (FEP) calculations: The contributions to the free energy difference from intraperturbed group interactions are negligible. We show: (A) that these contributions are not, in fact, necessarily negligible; and (B) that, in particular, the bond-stretching contribution resulting from changes in equilibrium bond lengths is important, but can be extremely difficult to calculate using standard procedures. We show how this contribution can be determined using an alternative "pseudopotential of mean force" correction. We also examine the behavior of the calculated free energy with the length of the simulation, and conclude that simulation times significantly longer than those typically used at present (e.g., greater-than-or-equal-to 200 ps) may be required to yield reliable and accurate results.