Adiabatic Free Energy Dynamics, introduced by Rosso et al. (J. Chem. Phys. 2002, 116, 4389) for computing free energy profiles quickly and accurately using adiabatic dynamics, has been used for such problems as conformational sampling of peptides (J. Phys. Chem. B 2005, 109, 4162) and computing alchemical free energies such as solvation free energy (J. Chem. Phys. 2006, 125, 074115). In both cases, the use of adiabatic dynamics has been shown to lead to significant speed increases versus traditional methods such as Umbrella Sampling, Thermodynamic Integration, and Free Energy Perturbation. The most significant drawback to the AFED approach is the need to transform the coordinate system into generalized coordinates.

Here we present an extension of the AFED method that circumvents such transformations. In this approach, additional degrees of freedom, s_i, are used to drive the collective variables of interest, s_i(r) via a harmonic potential coupling. The driving variables are given temperature sufficient to cross any significant barriers and mass sufficient to maintain adiabatic decoupling. The free energy surface along the collective variables is then constructed using the adiabatic probability distribution of the variables, s_i. This driven AFED approach (d-AFED) is employed to compare the conformational preferences of small peptides for three force fields. The results show that d-AFED accurately and efficiently computes the free energy surfaces of interest using radius of gyration, R_G, and number of hydrogen bonds, N_H, as collective variables.