Re: [AMBER] How can I get a average structure?

From: Jason Swails <>
Date: Tue, 28 Aug 2012 10:13:18 -0400

On Tue, Aug 28, 2012 at 10:05 AM, Sanjib Paul <>wrote:

> Dear Amber Users,
> I wanted to generate a average structure form 1
> ns trajectory file of my protein using 'average' command of ptraj. My input
> file is like following:
> trajin protein_prod_1ns.mdcrd
> average protein_prod_1ns_avg.pdb pdb

Note, the 'average' command does not do any kind of implicit fitting, so
you are including free rotation and translation in your average (which is
*NOT* something you probably want to do).

To get rid of these contributions, add a "rms" command before 'average'.

> When I observe my pdb file of average structure I noticed that some portion
> of protein is very unstable and very absurd. Atoms are overlapping and some
> unusual bonds are also present. Then, I followed analysis portion of
> TUTORIAL B3. Here also structure is like before. Some unusual bonds and
> overlapped atoms are present. How can I overcome it? Please help.

'Warped' and 'strange' structures are expected for an average (to some
extent). Consider for instance a freely rotating methyl group such that
all hydrogens attached to the CH3 are rotationally degenerate. In this
case, each hydrogen atom should have the *same* average coordinates which
should lie almost directly on top of the C atom they are all attached to.

You should think about what you really want and whether or not an 'average'
structure satisfies your needs.


Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Tue Aug 28 2012 - 07:30:04 PDT
Custom Search