Re: [AMBER] diff. between avg/rep structures & connectivity problem in PDB files

From: Daniel Roe <>
Date: Mon, 8 Oct 2012 08:39:42 -0600


On Sun, Oct 7, 2012 at 3:24 AM, vaibhav dixit <> wrote:
> If I guessed correctly, average structure means "average of all the
> coordinates" in a given cluster. So the avg. structure *might not be
> chemically meaningful* structure

Yes, the average structure is made up of the coordinate positions
averaged over all frames of the cluster, and is almost certainly not
chemically meaningful. However, it can be useful for using as a
reference structure for fluctuation analysis etc.

> How the representative structure is chosen by the clustering algorithm? No
> information is given in the manual and the original paper.

>From the paper: "The representative structure is the structure which
has the minimal sum of the squared displacements between other
structures in the cluster and itself."

> I'm having problem visualizing average PDB files generated by the
> clustering algorithm.
> Unusual bonds are appearing (especially in the side chains) when I load
> these file either in pymol or vmd.

Try doing output in (or alternatively converting your existing avg
PDBs to) Amber restart format and view them that way. Since Amber
topology contains bond info no spurious bonds will be created from
averaged coords.

Hope this is helpful.


Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 201
Salt Lake City, UT 84112-5820
(801) 587-9652
(801) 585-9119 (Fax)
AMBER mailing list
Received on Mon Oct 08 2012 - 08:00:03 PDT
Custom Search