[gmx-users] g_cluster -dm option

[Francesco Oteri]

Hi santhu,
maybe the differecence you observed depend by the xpm format.
In the xpm each cell is represented by one character that, in turn, is 
represented as a color by xpm2ps.
Therefore the character count cannot be infinite so each character can 
represent a range of RMSD cell.
In my opinion this rapresentation is the source of the problem.

You should modify g_cluster to write the RMSD matrix in a binary format 
to avoid rounding error and,
at the same  time, to save computational time.

Il 29/09/2011 22:25, santhu kumar ha scritto:
> Hello,
>
> I am using g_cluster to cluster trajectories.
> The exact command I am using is : g_cluster -s protein.pdb -f all.trr 
> -n index.ndx -o rmsd-clust.xpm -method gromos -cutoff 0.07 -wcl 250. 
> This command gives me 145 clusters.
>
> This command writes out a rmsd-clust.xpm, which is the RMSD distance 
> matrix. Since the trajectory is quite huge, i want to perform the RMSD 
> matrix computation only once.
> If I want to change options of clustering, I have tried giving the 
> computed matrix as input to another run but the results are quite 
> different.
>
> For example : g_cluster -s ubqs.pdb -f all.trr -n index.ndx -dm 
> rmsd-clust.xpm -method gromos -cutoff 0.07 -wcl 250 would then give 
> only 3 clusters.
>
> Can somebody please advice me on how to avoid recomputing the RMSD 
> distance matrix for each change in clustering option??
>
> Thanks
> Santhosh
>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110930/f560ef7e/attachment.html>