Re: [AMBER] about iwrap=1 and ntxo=2

From: Jason Swails <>
Date: Tue, 5 Feb 2013 15:27:48 -0500

On Sat, Feb 2, 2013 at 3:08 PM, Thomas Evangelidis <>wrote:

> I have two more questions:
> 1. Does the combination iwrap=0, ntxo=2, ioutfm=1 yield performance loss
> compared with iwrap=1, ntxo=2, ioutfm=1 ?

No performance loss. Minimum image conventions need to be satisfied
regardless of how the 'master' set of coordinates are imaged.

> 2. In an older post Prof Case recommended avoiding the usage of NMR-based
> restraints (nmropt=1) with iwrap=1. Does the same apply to ntr=1, iwrap=1,
> nmropt=0? A part of my protein sticks out of the unit cell although I have
> applied coordinate restraints to 20 residues (restraint_wt=200) - there's a
> slight drift to the whole protein in spite of the strong coordinate
> restraints.

The main thing that will cause problems is if one of your molecules that
you are constraining is 'wrapped' to the other side of the unit cell. If
this happens, then the reference coordinates and the actual coordinates
will differ by a box length, which will likely blow up your simulation.

There are a few reasons this is unlikely, though: first, if you have
imposed positional restraints, even 'weak' ones between 0.1 and 1
kcal/mol/A/A, that molecule will be unlikely to move far enough away from
the center of the restraint to be wrapped in the first place. Secondly,
the main molecules that get wrapped are waters (and these are rarely
restrained). Large molecules with restraints on a large number of atoms
(e.g., backbone restraints on a protein) will move so little that there is
no chance that the COM of that molecule will move "outside" the current
unit cell.

There is a small chance that a _very_ weakly restrained, unbound ligand
will start out close enough to the box edge that it will be imaged, thereby
blowing up your simulation, but this is unlikely.

The main concern with the NMR restraints, though, is that they are far more
often spread over long distances, between different molecules, which could
lead to a 'reimaging' disaster. Intra-molecular restraints with nmropt=1
are still immune to imaging artifacts (since imaging is done on a
per-molecule basis).

As for the drift, this may be due to COM translation removal (see the nscm
variable). Try setting nscm larger than nstlim and see if the restrained
atoms still drift.

Using ntxo=2 and ioutfm=1, though, there's no risk of coordinate overflow,
so you might as well set iwrap=0 and avoid these worries (you can always
use cpptraj/ptraj to reimage afterwards).


Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Tue Feb 05 2013 - 12:30:02 PST
Custom Search