A schematic illustration of the ResEx method with two replicas, one at the atomistic (AA) resolution and one at the CG resolution. The system DOFs are divided into CG DOF, denoted by capital letters , and the AA DOF, denoted by lower-case letters . The configuration exchange between AA and CG replicas is attempted periodically. If successful, the extra DOFs for the AA replica, , are dropped and those for the CG replica, , are regenerated. The swapping step of ResEx can be considered as a walk in an extended configuration space from the old state to the new state .
A schematic illustration of the CBMC. In this scheme, the molecular structure is built one site at a time using a look-ahead algorithm. (a) Given the positions of monomeric units , , and , in order to add monomeric unit, a set of trial positions are generated. Denoting the energy associated with the trial position as , we assign a weight . (b) One trial position is selected according to its Rosenbluth factor, . (c) Other monomeric units can be sequentially grown in a similar fashion.
The connectivity for the known atoms A, B, C, and the hydrogen atoms , , to be regrown for a group.
(a) The atomistic (opaque ball-and-stick representation) and CG models (transparent VDW ball representation) for the butane molecules. (b) The exchange history between AA and CG models. The top and bottom parallel lines correspond to the CG and atomistic replicas, respectively. The vertical lines represent the successful configuration exchanges between these two replicas. For clarity, only the history of the first was plotted. (c) The comparison of the C–C–C–C dihedral angle distribution between a CB-ResEx (circle) and that from a REMD (solid line) simulation. The corresponding distribution of CG replica in CB-ResEx (square) is also plotted.
The logarithm of the logarithmic population distributions of (the angle between ) and (the angle between ) backbone dihedral angles of dialanine from REMD simulation (top panel) and CB-ResEx (bottom panel) simulation. The values are color coded from blue (high population) to red (low population).
The running estimate of the population difference between two important states and of a dialanine system as a function of the simulation time. The dashed lines are the results from four independent the standard CTMD simulation. The solid line with solid circles is the running average of four CB-ResEx simulations, and the corresponding error bars represent the maximum negative and positive deviations from the average value. The simulation time does not include any computational overhead associated with the CB-ResEx. These results show that CB-ResEx method has no improvement in conformation space sampling over the CTMD simulation. The underlying reason is discussed in the text.
The trajectories of an important backbone dihedral angle for the CTMD (top panel) and the CB-ResEx (bottom panel) simulation for alaine dipeptide with the constraint potential.
The running estimate of the population ratios between two important states and of a dialanine system with the biased potential [Eq. (10)]. The dashed lines are the results from eight independent CTMD simulation. The solid line with solid circles is the running average of eight CB-ResEx simulations, and the corresponding error bars represent the maximum negative and positive deviations from the average value. The simulation time does not include any computational overhead associated with trial move generation for CB-ResEx. These results show that CB-ResEx has significant improvement in configurational sampling over the CTMD simulation, which is consistent with the high transition frequency shown in Fig. 7.
Article metrics loading...
Full text loading...