Re: AMBER: distance between images

From: Thomas E. Cheatham, III <>
Date: Wed, 5 Jan 2005 10:29:19 -0700 (Mountain Standard Time)

> I would like to analyse the minimum distance between my solute and its images
> during my simulation. I am working with an octahedron box of solvent.

The distance command in ptraj will calculate the minimum imaged distance
by default, unless you specify the "noimage" keyword. This is a bit
computationally demanding if you are in a non-orthorhombic box since
rather than be clever, all 26 possible imaged distances are calculated and
the shortest reported. This command is really useful to see if you are
interacting closely with your periodic image (as I've seen in DNA where in
a small box the DNA lined up with its images to form an almost infinite

        distance d1 out d1.dat :1 :10

will calculate the minimum imaged distance from the center of mass of
residue 1 to that of residue 10. IF you set prnlev > 3, you can see all
the distances. Note that almost all of the distance measures in ptraj use
the minimum image (such as with h-bonding) except for diffusion (at

Note that this doesn't report the shortest distance between the molecule
and its image, but between different groups. To find the minimum overall
would require ~ N**2 * 26 measurements per frame (i.e. distance of every
atom to every (other) atom, imaged) which would be very costly. A hellish
ptraj script could be written to provide the measures (probably on a
residue to residue basis).

Another possible way (speculating here) to get this info would be to do a
self radial distribution function with and without imaging; where the two
plots differed would show close approach of the images and the distance,
although it wouldn't identify which were the close atoms. For this to
work, the entire molecule would have to fit within the box and be centered
at the origin before doing the noimage calculation...

radial radial_noimage 0.5 12.0 .CA @CA noimage


radial radial_image 0.5 12.0 .CA @CA

Good luck.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Jan 05 2005 - 17:53:00 PST
Custom Search