Metrics used for comparison of binding site conformations
The following metrics can be chosen:
- Back-bone atoms: RMSD of back-bone atoms for each residue from their position in the reference structure are computed.
- Geometric center of each residue: Absolute values of the deviations in position of geometric centers of residues (taking into account only heavy atoms) in the binding site from those of the reference structure are computed.
By default, the back-bone atoms metric is used.
- Snapshots of a particular trajectory are aligned and superimposed with the reference structure using binding site residues and the chosen metrics, herewith for each snapshot in a trajectory, the RMSD from the reference structure is computed;
- Computed RMSD matrix (residue vs snapshot), as well as averaged RMSD for each snapshot over all residues and maximum RMSD of each residue over all snapshots, are visualized using Matplotlib library and included in the output HTML page (see TRAPP-analysis output).
- Simplified single-linkage hierarchical clustering is carried out:
- K-means clustering is carried out:
To assign a structure to a particular cluster, a comparison with only the first element of each cluster is required.
To define the number of clusters, distances between k-means centers of clusters obtained during single-linkage hierarchical clustering are compared pairwise and clusters that have distances smaller than the cluster threshold are merged. New centers are then computed and used as a start for k-means clustering.