While atomistic simulations allow for the study of protein folding in atomic detail, it has been difficult to connect simulations with experimental measurements due to the fact that simulations typically consider the dynamics of a single molecule over short timescales
(nanoseconds) while high-resolution experiments such as laser T-jump monitor the time evolution of an ensemble of molecules over long timescales (microseconds). To bridge this gap, as well as extract insight about statistical macromolecular dynamics over timescales much longer than are computationally accessible, we and others have proposed the construction of discrete-state master equation or Markov models from many short trajectories which would describe statistical dynamics over long timescales. Constructing these models requires both a method of identifying kinetically metastable states, a process tantamount to identifying the slow degrees of freedom, and an efficient method for computing transition probabilities or rates between the states. We present our recent progress toward the automatic construction of these models, validation that they accurately reproduce the kinetic behavior of the simulations from which they were constructed, and direct comparison with laser T-jump experiments.
Back to Rare Events in High-Dimensional Systems