Figure 9.

Markov state modeling. (a) Free energy surface (FES) projected onto the first two independent components of TICA (TICs) at a TICA lag time of 750 ns. The centers of five metastable states obtained by PCCA+ decomposition are superimposed on the contour plot. The area of each circle represents its equilibrium population. (b) Network representation of the MSM (lag time of 250 ns) coarse-grained onto the five metastable states. The population of each state (π) is indicated in the figure. The state number 1 (red) is the state to which all the starting frames of the trajectories have been assigned, thus we consider this macrostate to be the crystal structure-encompassing state. The state number 4 (cyan) is the state with the highest population, thus we label it as dominant state. c) Transition Path Theory (TPT) analysis of the loop dynamics, i.e., the ensemble of reactive pathways connecting the crystal structure-encompassing and the dominant state. Reactive transitions probabilities (fluxes) are shown as rates (probability per time unit) in μs⁻¹. The guanine strands connected by the loops are shown in gray lines. The ensemble of loop backbone geometries contained in each state is shown by displaying overlays of the most probable structures of the state (opaque lines) on top of samples of the entire state (transparent lines) to show both the intrastate conformational variability and the interstate conformational differences. The most probable position of loop nucleotides in each state is shown in licorice; PT1, PT2, and PA3 are shown in green, yellow, and blue, respectively.