This program performs clustering based on a given radius using the same algorithm as in the CHARMM cluster module. In the MMTSB Tool Set it is used by the Cluster.pm package as an alternative to jclust but can also be called separately.
Most command line options are shared with jclust. The following cluster modes are available: rmsd: based on cartesian coordinate RMSD, phi: based on phi backbone dihedral angle RMSD, psi: based on psi backbone dihedral angle RMSD, phipsi: based on the sum of phi and psi RMSD values divided by two, mix (followed by a mixing factor): based on the sum of phi and psi RMSD values divided by 20 and multiplied by the mixing factor plus the cartesian RMSD multiplied by 1 minus the mixing factor. Contact map based clustering is not available and the corresponding options are not recognized.
The cluster radius is given with -radius. The algorithm used in CHARMM finds an iterative solution from the initial cluster assignment until cluster centers differ by less than a given error tolerance. By default, kclust does not iterate and returns the initial clustering result. With the option -iterate the cluster result is iterated as in CHARMM using an error tolerance given with -maxerr.
Cluster centroids are written out if the option -centroid is given. The option -cdist prints out the distance of each cluster member to the centroid.