Hi,
Are you talking about the "test-out.rep.cX" files, "test-out.c*"
files, or "test-out.avg.c*" files? If you are talking about the
"test-out.avg.c*" files this behavior is not unexpected since you are
averaging over the coordinates of all structures within that cluster.
In my opinion the average structure isn't really that useful (I never
use the average keyword), since it does tend to be distorted. I think
it's better to just calculate whatever property you're looking for (a
distance, RMS, etc) on the entire cluster ("test.out.c*") and average
that.
Also, why are you centering and imaging your coordinates at all? Since
the cluster command you are using uses best-fit RMSD as a metric it
doesn't matter where your coordinates start from. Just do the
clustering and forget about imaging and centering - it won't change
the results and will be faster.
Hope this is helpful.
-Dan
On Sat, Aug 20, 2011 at 7:08 AM, George Tzotzos <gtzotzos.me.com> wrote:
> Hi everybody,
>
> I'm trying to cluster the trajectory frames of a 2ns production run of a solvated complex (138 residues).
>
> Using ptrja my.prmtop cluster.in
>
> where cluster.in is
>
> trajin prod_2ns.mdcrd
> center :1-138 mass origin
> image origin center familiar
> cluster out test-out representative pdb \
> average pdb averagelinkage clusters 5.0 mass rms :1-138 .CA
>
> produces clusters. However, when I try to visualise the clusters some frames show with unrealistic bond lengths.
>
> I believe the problems is in the centering and imaging.
>
> I'd appreciate if anyone could help to correct the above script.
>
> Thanks in advance
>
> George
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Aug 22 2011 - 09:30:03 PDT
