This program performs hierarchical clustering on a set of files. In the MMTSB Tool Set it is used by the Cluster.pm package but can also be called separately.
The command line parameters consist of a number of options and a file name containing a list of files for which the clustering algorithm should be applied. If no file name is given the file names are read from standard input.
The following options are available: -mode select the clustering mode (rmsd: RMSD based or contact: contact map based). The maximum residue distance up to which residue contacts are considered in the comparison can be changed from the default value of 12 A with -maxdist. For RMSD-based clustering the options -ca, -cb, -cab, -heavy, and -all are used to specify which atoms are used for the RMSD calculation. The option -lsqfit requests a least squares fit before RMSD values are calculated. For fragment/loop modeling a residue subset may be given with -l and -fitxl selects that the proteins structures sourrounding the given residue subset is fit rather than the residues themselves. Alternatively a different residue subset used only for fitting may also be given with the -fit option.
The format of the input files can be either PDB format (selected with -pdb) or SICHO chain format (-sicho). A maximum number of clusters can be set with -maxnum. Otherwise the maximum number is set to half of the number of input structures.
By default the output does not contain the centroids for each cluster. This may be requested with -centroid. Finally, -ref is available to provide a reference RGB map which is used for writing out centroid in contact-based clustering as a difference map with respect to that reference.
An alternative clustering program is available in kclust.