Examining the limits of time reweighting and Kramers' rate theory to obtain correct kinetics from accelerated molecular dynamics

Accelerated molecular dynamics simulations are routinely being used to recover the correct canonical probability distributions corresponding to the original potential energy landscape of biomolecular systems. However, the limits of time reweighting, based on transition state theory, in obtaining true kinetic rates from accelerated molecular dynamics for biomolecular systems are less obvious. Here, we investigate this issue by studying the kinetics of cis-trans isomerization of peptidie omega bond by accelerated molecular dynamics. We find that time reweighting is valid for obtaining true kinetics when the original potential is not altered at the transition state regions, as expected. When the original potential landscape is modified such that the applied boost potential alters the transition state regions, time reweighting fails to reproduce correct kinetics and the reweighted rate is much slower than the true rate. By adopting the overdamped limit of Kramers' rate theory, we are successful in recovering correct kinetics irrespective of whether or not the transition state regions are modified. Furthermore, we tested the validity of the acceleration weight factor from the path integral formalism for obtaining the correct kinetics of cis-trans isomerization. It was found that this formulation of the weight factor is not suitable for long time scale processes such as cis-trans isomerization with high energy barriers. (C) 2010 American Institute of Physics. [doi:10.1063/1.3432761]