Finding number of structures per cluster in rna_cluster

[mandar5335]

Hi,

Is there any way to find number of structures per cluster after using rna_cluster ?

I can extract cluster representative using extract_pdbs, but could not get much more information about each cluster.

Best,
Mandar

[roccomoretti]

In the silent file output, there's a nclust entry which I believe should be the total number of poses in the corresponding cluster. This may or may not make its way to the information at the bottom of the PDB file when you run extract_pdbs.

If not, you can always grep the silent file for the SCORE: line with the corresponding structure tag, and then attempt to extract the nclust value from the appropriate position in the score line.

[mandar5335]

Thanks for the reply, I could get nclust values.

However, I am facing another issue with rna_cluster. I am using rna_cluster to cluster top 1% score sorted structures with 5 Angstrom radius.
I am assuming rna_cluster sorts by score structures and then clusters top 167 score-sorted structures.

rna_cluster.mpi.linuxgccrelease -in:file:silent ../simple_cycle.out -out:file:silent clustered.out -nstruct 167 -cluster:radius 5

However, all 167 structures are assigned to single cluster. I see RMSD difference is ~ 16 Angstroms in top 2 structures.

For another Rosetta run, also, I observed similar result. There I get different clusters. But RMSD (calculated with Pymol, rechecked with Gromacs's gmx rms) is more than 5 A for top 2 structures in the same cluster.

Is there anything missing in my command line?

[mandar5335]

Hi,
RMSD of structures in same cluster is higher than cutoff used for clustering. Any suggestions on rna_cluster RMSD clustering?