Rational choice of molecular dynamics simulation parameters through the use of the three-dimensional autocorrelation method: application to calmodulin flexibility study.

Abstract

We examined the effects of several adjustable parameters for use in molecular dynamics simulations of proteins using both standard criteria (radius of gyration, root mean square deviation from starting coordinates, molecular mechanics energy) and a new description of protein conformations by 3-D autocorrelation vectors (3-D ACV). We chose calmodulin (CaM) as a protein model and analysed 23 simulations using different combinations of the four molecular dynamics parameters studied, such as the dielectric constant (epsilon), the heating phase time (H), the thermal bath coupling time (zeta T) and the time step size (delta t). The correctness of the various trajectories generated with different parameter sets was evaluated through geometric analysis and use of a knowledge-based profile method. It is shown that 3-D ACV combined with multivariate statistical analysis provides a convenient way to describe and compare molecular dynamics simulations and constitutes a valuable complementary tool to standard methods. Using these methods, comparison of the various simulations performed on CaM indicated that the best in vacuo parameter set was epsilon = 1 x r, H = 15 ps, zeta T = 0.1 ps and delta t = 1 fs in fairly good agreement with previous less extensive comparisons of molecular dynamics trajectories.