Re: [AMBER] restraints on group of atoms

From: David A Case <>
Date: Mon, 14 Dec 2015 08:06:57 -0500

On Mon, Dec 14, 2015, Dhananjay wrote:
> I want to keep the two chins at certain distance limit such that the centre
> of geometry (CoG) distance difference will be between 50.0 to 60.0
> Angstroms
> (i.e., I want 50.0 <= [CoG(chainA)-CoG(chianB)] <= 60.0 )
> To achieve this, I specify: r1=50.0, r2=55.0, r3=55.0, r4=60.0, rk2=10.0,
> rk3=10.0,

This might not do what you want it to; you should plot the restraint
function vs distance, and see the sort of behavior you get with various
choices of r1, r2, r3, r4.

>From your verbal description, I would use r1=30, r2=50, r3=60 and r4=80, then
try to choose rk2 and rk3 so that the steepness of the penalty outside the
50--60 range is enough to restrain the system.

> Though the chin A and B do not exceed 60.0 angstroms, i.e. upper limit,
> however, quite often fall close with distance less than 50.0 angstroms
> (~30.0 Ang.).

This could be because the system really wants to go to short distances; take
some snapshots at 30 Ang. and look at the restraint energy and system energy;
compare those to typical results where the distance is above 50 Ang. See if
you can learn what sort of forces you can have in your system that would favor
short distances.

> I am using Amber 14 with pmemd.cuda version on GPU with all latest updates.

Be *sure* to compare sample results between GPU and CPU runs: support for
these sorts of restraints in the GPU code is rather recent, and you might be
encountering a bug.


AMBER mailing list
Received on Mon Dec 14 2015 - 05:30:03 PST
Custom Search