[gmx-users] cluster analysis on HREX results

[Shyno Mathew]

Dear all,

1. For the cluster analysis, I am using the gromos method with cutoff 0.25
nm.
With this cutoff I got 201 structures for the lowest temperature for
trajectories that are continuous in time (after using demux.pl)
The rmsd ranges from 0.073 to 1.55nm.

If I reduce the cutoff to 0.1 nm I will get 1919 structures, which is very
close to the total number of frames, 2001 in the .xtc file. This doesn't
make sense as I am getting one structure for each frame. Here 1 frame
corresponds to 0.01 ns

My question, with the cluster analysis, do you fix the value of cutoff or
you change cutoff to get a certain number of structures?

Following is the syntax I used:
 g_cluster_mpi -s PRO0.gro -f redPRO0.xtc -method gromos -sz clustsize0.xvg
-o clusters0.xpm -g cluster0.log -clid clusidovertime0.xvg -cl out0.pdb
-cutoff 0.25

2. For replica exchange,is it better to do the cluster analysis on
trajectories that are continuous in ensemble or trajectories that are
continuous in time (after using demux.pl)
I tried both ways:
For replica 0 (for trajectories that are continuous in ensemble, at lowest
temperature 300 K), I got two structures with cutoff 0.25
For replica0 (for trajectories that are continuous in time), I got 201
structures with cutoff 0.25

Thanks in advance for your help.

thanks,
Shyno
-- 
Shyno Mathew
PhD Candidate
Department of Chemical Engineering
Graduate Assistant
Office of Graduate Student Affairs
The Fu Foundation School Of Engineering and Applied Science
Columbia University