rna_cluster issue: Structures within same cluster have RMSD higher than cutoff

[mandar5335]

Hi,

I am using rna_cluster to cluster top 1% of structures after FARFAR2 run.

rna_cluster.mpi.linuxgccrelease -in:file:silent ../simple_cycle.out -out:file:silent clustered.out -nstruct 2000 -cluster:radius 5

After that structures from same cluster (confirmed with nclust id) were extracted from clustered.out with extract_pdbs command. However, RMSD between these structures is higher than 5 Angstrom (~15 Angstrom).

This was confirmed with Pymol, Gromac's gmx rms command as well.

Is this bug or Is something missing in rna_cluster command line?

Best,
Mandar

[mandar5335]

Update on question:

1. Assuming it could be MPI version issue, I tried serial rna_cluster.default.linuxgccrelease version. However, result remains same.

2. I have attached histogram of pairwise of RMSD of all structures in Cluster 1. median RMSD is ~ 12.5 Angstroms, far away from cutoff of 5 Angstroms.
   

It would be helpful to receive comment from experts as rna_cluster command is straightforward and I am not sure whether result is a bug or anything trivial part is missing.

TIA,
Mandar