AMBER Archive (2008)

Subject: Re: AMBER: problem with cartesian restraints with NTP simulations

From: David A. Case (
Date: Wed Nov 19 2008 - 09:02:56 CST

On Wed, Nov 19, 2008, Barbault Florent wrote:
> I am facing a problem with restraints on Amber 9. I am using cartesian
> restraints (ntr=1) with this sort of mask:
> restraintmask = '(:112 >@8.0) :1-111!@H'
> I tried to do some MD with this sort of restraint in explicit solvent.
> Generally, I share my simulation in several runs (generaly 100ps). When I do
> these simulations in NVT I have no problem (ntb=1). However, if I launch the
> same simulation in NTP, my restraint energy always begins with a huge value
> on every starting runs. I don't have this problem if I don't share my
> trajectory into several runs. Therefore, I think my problem arises from the
> reading of coordinates when the non-bonded atom list is created. Is there any
> people who have an idea how to solve my problem?

You don't say what you are using for your starting (-c) and reference (-ref)
coordinates, but I'm guessing(?) that these are not the same, and that the box
size in the NPT simulation coordinates changes (in multiple runs) but that you
are still using reference coordinates with a box size from the original
system. That would explain the behavior you see.

[It might also be a wrapping problem, where the simulated system gets wrapped
but the reference coordinates stay the same. You might try setting iwrap=1 to
see if that helps. But it is not obvious why this would happen only with NPT
turned on, unless it is just a coincidence.]

If you really want to always restrain to a particular set of initial
coordinates, there is no simple fix. You could carefully equilibrate with NTP
(say keeping the solute coordinates restrained the the position you want), and
then use those coordinates as a starting point for NVT simulations -- unless
you have large conformational changes, the volume won't change much after a
good equilibration.

Other people on the list may have some more clever ideas here.


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