What is eBDIMS?
eBDIMS (elastic-network driven Brownian Dynamics Importance Sampling) is a coarse-grained path sampling method. It generates feasible transition pathways between two protein conformations by running a simplified simulation where proteins are modelled as an elastic network.
As described in the original research article, eBDIMS is particularly powerful in combination with projection analysis in simplified “motion spaces” defined by the major movements explored by the protein of interest.
Specifically, principal component analysis (PCA) of structural information available in the Protein Data Bank (PDB) for a given protein can provide "motion" axes useful to understand how a conformational transition happens.
Projection of eBDIMS pathways onto the PC-space can help to classify structures, identify on-pathway intermediates, or monitor the sampling by Molecular Dynamics simulations (Figure 1).
How is eBDIMS different from other path-sampling methods?
The eBDIMS algorithm has several unique features:
- Experimental validation: eBDIMS has been shown to accurately and spontaneously predict transition intermediates between different conformations which have been trapped experimentally (Orellana et al., 2016). This is done without any previous information on their structure.
- Transitions explore “natural” motions: instead of a straight line (like morphing methods) or zig-zag trajectories (usual in many path-sampling algorithms) eBDIMS generates pathways that follow the “natural” motions of proteins observed experimentally in structural ensembles. The pathways not only find intermediates, but do this by following "natural" motions.
- Definition of a low energy region: In contrast to typical morphing or path-sampling methods which generate a single linear trajectory, eBDIMS generates two different non-linear transitions that connect stable end-states (e.g. from open to closed and vice versa). These two transitions define the borders of the low-energy area for the conformational change sampled by complex MD simulations and experimental structures.
What is projection analysis useful for?
Grasping the similarities and differences between multiple conformations of the same protein is a difficult task, which normally requires arbitrary definition of variables to distinguish them from each other (e.g. distance between certain residues, orientation of a helix, diameter of a cavity, etc).
Principal component analysis can extract the dominant motions or Principal Components (PCs) contained in a structural ensemble, providing natural variables for structure comparison. Projections onto PC-motion space are like a visual road map of the conformational space, which allows for immediate classification of structure clusters and how they relate to transition paths, sampling by simulations, etc.
What are the eBDIMS server applications?
The eBDIMS server generates feasible paths between start and target structures and allows for their visualization in a 2D-motion space.
Case 1: Pathway generation
Input: two different conformations (start and target)
If two end-states structures are provided, the server simply generates two trajectories that connect them.
Projections are generated onto Normal Modes to allow for a quick inspection of the pathways in order to evaluate their asymmetry or smoothness.
Case 2: Pathway generation and ensemble analysis
Input: two different conformations (start and target) plus additional structures (e.g.potential intermediates, simulations, etc).
When several conformations are available (from experiment or simulation), the server can additionally compute their PCs to generate a 2D- structural landscape. This allows for automatic clustering of the structures, immediate visual evaluation of the generated paths and their relationship with multiple conformations.
The eBDIMS server runs with default parameters. The eBDIMS stand-alone package allows to run the path-generation algorithm modyfing these generic values, to generate slightly different paths.