Free energy derivatives: A new method for probing the convergence problem in free energy calculations

The convergence behavior of free energy calculations has been explored in more detail than in any previously reported work, using a model system of two neon atoms in a periodic box of water. We find that for thermodynamic integration‐type free energy calculations as much as a nanosecond or more molecular dynamics sampling is required to obtain a fully converged value for a single λ point of the integrand. The concept of “free energy derivatives” with respect to the individual parameters of the force field is introduced. This formalism allows the total convergence of the simulation to be deconvoluted into components. A determination of the statistical “sampling ratio” from these simulations indicates that for window‐type free energy calculations carried out in a periodic waterbox of typical size at least 0.6 ps of sampling should be performed at each window (0.7 ps if constraint contributions to the free energy are being determined). General methods to estimate and reduce the error in thermodynamic integration and free energy perturbation calculations are discussed. We show that the difficulty in applying such methods is determining a reliable estimate of the correlation length from a short series of data. © 1994 by John Wiley & Sons, Inc.

[1]  P. A. Bash,et al.  Free energy calculations by computer simulation. , 1987, Science.

[2]  Elvira Guàrdia,et al.  Na+–Na+ and Cl−–Cl− ion pairs in water: Mean force potentials by constrained molecular dynamics , 1991 .

[3]  Kyoko Watanabe,et al.  Molecular dynamics study of the hydrophobic interaction in an aqueous solution of krypton , 1986 .

[4]  Robert H. Wood,et al.  Estimation of errors in free energy calculations due to the lag between the hamiltonian and the system configuration , 1991 .

[5]  Alfons Geiger,et al.  Molecular dynamics study of the hydration of Lennard‐Jones solutes , 1979 .

[6]  B. H. Wells,et al.  Estimating errors in molecular simulation calculations , 1984 .

[7]  P. Kollman,et al.  An all atom force field for simulations of proteins and nucleic acids , 1986, Journal of computational chemistry.

[8]  Chung F. Wong,et al.  Sensitivity analysis and principal component analysis in free energy calculations , 1991 .

[9]  D. Wallace,et al.  Statistical errors in molecular dynamics averages , 1985 .

[10]  David E. Smith,et al.  Entropy of association of methane in water : a new molecular dynamics computer simulation , 1992 .

[11]  W. L. Jorgensen,et al.  Comparison of simple potential functions for simulating liquid water , 1983 .

[12]  Peter A. Kollman,et al.  Computer modeling of the interactions of complex molecules , 1990 .

[13]  J. Andrew McCammon,et al.  Free energy difference calculations by thermodynamic integration: Difficulties in obtaining a precise value , 1991 .

[14]  Hideaki Umeyama,et al.  Free energy perturbation calculations on multiple mutation bases , 1992 .

[15]  Peter A. Kollman,et al.  The lag between the Hamiltonian and the system configuration in free energy perturbation calculations , 1989 .

[16]  David A. Pearlman,et al.  Determining the contributions of constraints in free energy calculations: Development, characterization, and recommendations , 1993 .

[17]  William L. Jorgensen,et al.  Theoretical studies of medium effects on conformational equilibria , 1983 .

[18]  Robert H. Wood,et al.  Systematic errors in free energy perturbation calculations due to a finite sample of configuration space: sample-size hysteresis , 1991 .

[19]  Peter A. Kollman,et al.  A new method for carrying out free energy perturbation calculations: Dynamically modified windows , 1989 .

[20]  H. Berendsen,et al.  THERMODYNAMICS OF CAVITY FORMATION IN WATER - A MOLECULAR-DYNAMICS STUDY , 1982 .

[21]  A. J. Stam,et al.  Estimation of statistical errors in molecular simulation calculations , 1986 .

[22]  T. Straatsma,et al.  Multiconfiguration thermodynamic integration , 1991 .

[23]  W. L. Jorgensen,et al.  Monte Carlo simulation of differences in free energies of hydration , 1985 .

[24]  Peter A. Kollman,et al.  The overlooked bond‐stretching contribution in free energy perturbation calculations , 1991 .

[25]  Alan E. Mark,et al.  Calculation of Relative Free-Energy Via Indirect Pathways , 1991 .

[26]  B. Montgomery Pettitt,et al.  Convergence of the Chemical Potential in Aqueous simulations , 1991 .

[27]  P A Kollman,et al.  The calculated free energy effects of 5‐methyl cytosine on the B to Z transition in DNA , 1990, Biopolymers.

[28]  Mihaly Mezei,et al.  Monte Carlo computer simulation study of the hydrophobic effect. Potential of mean force for [(CH4)2]aq at 25 and 50 °C , 1982 .

[29]  Wieslaw Nowak,et al.  Locally enhanced sampling in free energy calculations: Application of mean field approximation to accurate calculation of free energy differences , 1992 .

[30]  Jan Hermans,et al.  Simple analysis of noise and hysteresis in (slow-growth) free energy simulations , 1991 .

[31]  G. Ciccotti,et al.  Numerical Integration of the Cartesian Equations of Motion of a System with Constraints: Molecular Dynamics of n-Alkanes , 1977 .

[32]  Yunyu Shi,et al.  Calculations of relative free energy surfaces in configuration space using an integration method , 1991 .

[33]  M Karplus,et al.  Dynamics of proteins: elements and function. , 1983, Annual review of biochemistry.