AMBER Archive (2008)

Subject: Re: AMBER: Does clustering involve fitting?

From: Jianyin Shao (jyshao2004_at_gmail.com)
Date: Wed Jul 30 2008 - 16:29:57 CDT


Hi, Steve,

The clustering will re-fit the atoms in order to calculate the RMSD and the
cluster center. And there is no option to turn it off right now. Another
user also discuss this issue some days early. Maybe we could have this
option later.

Here are something you may give it a try.
1. If you think the majority of your protein is stable except the loop, you
can try to include some other residues in clustering. Say the loop is
:90-100, you can try clustering with :85-105 or some other residues
spatially close to the loop.
2. The SOM, Bayesian and Cobweb algorithms do not require the calculation of
cluster center during the clustering. You may fit the trajectory first and
then run clustering with these algorithms. But the final reporting clusters
will still be rms-fitted.

Good luck,

Jianyin Shao

On Fri, Jul 25, 2008 at 2:02 PM, Steve Spronk <spronk_at_umich.edu> wrote:

> Hello all,
>
>
>
> The clustering utility of ptraj involves an RMSD calculation. In
> performing this calculation, are the two structures fit in order to
> determine the RMSD, or are the coordinates just used as they are read in?
> I'm interested to know because we'd like to use the rms command to fit the
> whole protein, but then cluster the resulting structures (without
> re-fitting) by measuring the RMSD values of just a single loop. But I'm
> concerned that the cluster command itself will re-fit the structures by
> overlaying the atoms of the loop. If this re-fitting is the case, is there
> an option to turn it off? I'd appreciate some clarification on this issue.
>
>
>
>
> Thanks,
>
> Steve
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu