Binding-induced functional-domain motions in the Argonaute characterized by adaptive advanced sampling
Fig 1
Adaptive H-REMD simulation algorithm.
(a) The python library and the Amber MD package are initialized using a batch file that contains the input parameters. There is no biasing force inserted to the system in the first interval. At the end of each interval the trajectories are passed to the library and CVs are calculated. Based on their values in the previous runs the biasing force is updated. The H-REMD simulation then continues with the updated biasing potential introduced in the replicas. The cycle goes on until the simulation is stopped. (b) Illustration of the biasing potential acting on an exemplary CV vs. time. The value of CV is coloured based on the simulation interval. At the beginning the biasing along the CV is localized to the vicinity of the initially sampled regime eventually with sampling of new regions the biasing potential extends to other regions along the CV. The overall bias potential (solid black line) is the sum of three Gaussians (dashed grey lines). Note, that in the reference replica (replica 1) there is no biasing and the biasing increases incrementally with increasing replica number.