Sorry for the naive and long question, but I noticed that the clustering program from version 2.3 was very well documented but the 3.0 version is less so. Aso, when downloading V.2.3 as a bundle, there is a set of scripts that contain various clustering analysis tools to help process the 2.3 clustering output (i.e. make_color_trees.py to construct dendrograms etc.) However, there are no scripts in the newer 3.0 and 3.1 releases so I just assumed the python scripts are no longer needed to interpret the clustering output in order to choose several potential decoys with the lowest energy and > nearest neighbors etc. for further analysis after ab initio.
Briefly, I have generated 40K decoys and clustered them with cluster.linuxrelease.exe using the default settings with the exception of asking for the top 100 cluster pdbs ordered from lowest energy. The CLUSTER.log file has a histogram at the bottom of the file represented as row after row of what appear to be some type of ordered ratios. In addition 100 pdb files were written. After studying the top 1-10 pdb files with the lowest energy, I was not sure how to proceed. I guess I was looking for a way to see which structure cluster is the largest with the lowest energy etc, but can't locate that info in the log file (only the energy calc for each of the 100 pdb files are listed).
Am I correct in assuming that since this version of CLUSTER is the newest that maybe the top 10 cluster centers (pdbs ?) from this list of 100 output .pdb files are representative or at least as close as I'm going to get to the global energy minimum (after idealizing etc) ?