Re: [AMBER] Is It Possible to Prevent Center-of-Mass Rotation During MD?

From: Gustavo Seabra <>
Date: Thu, 8 Jul 2010 09:30:31 -0300

Another option would be to use solvateoct instead of solvatebox, to
generate an octahedral box (take a look into AmberTools manual for
details). Still, you'd need to estimate the maximum size of your
system and calculate the padding from there.


On Thu, Jul 8, 2010 at 1:15 AM, Jason Swails <> wrote:
> Hello,
> I've put my comments below in line with your email.
> On Wed, Jul 7, 2010 at 10:41 PM, Logan Ahlstrom <>wrote:
>> In regards to the issue stated below, is there any way to prevent the
>> center-of-mass (COM) rotation of a molecule during simulation?  I have
>> come-across the nscm flag, but this will only remove translational COM for
>> periodic simulations. Any suggestions would be much appreciated.
> As far as I know, there is no way of removing rotations during the
> simulation.  The only way I know of doing this is to use ptraj to align the
> trajectory to a reference, the first structure, the last structure, etc.
> after the trajectory has been created.
>> Sincerely,
>> Logan S. Ahlstrom
>> University of Arizona
>> Department of Chemistry and Biochemistry
>> Miyashita Lab
>> I am currently running an explicit water REMD simulation of a dimer that
>> roughly has the shape of a “dumbbell”. Thus, when I solvate the system by
>> simply specifying a padding distance extending from the molecular
>> dimensions, a rectangle results.  However, during simulation the molecule
>> rotates such that the long side of the “dumbbell” extends along the
>> shortest
>> side of the rectangle and the protein becomes too close to its periodic
>> neighbors (i.e. within the the van der Waals and electrostatic cut-off I
>> have specified).  As for now, the only way I can think to fix this problem
>> is to solvate my system in a large cube such that the length of each side
>> is
>> greater than the length of the “dumbbell” plus the cut-off distance.
>> However, this dramatically increases the number of water molecules needed
>> for simulation.
> Even in this situation, if there was a way to remove rotational motions, it
> would seem to me that the rotational motions of the entire system (including
> water) would have to be eliminated, or you'd have, in all likelihood, van
> der waals and electrostatic clashes as the protein was rotated (wrt the
> frame of reference of the water box) onto a water molecule.  Therefore, even
> if an nscm for rotations did exist, I don't think it would help much in this
> situation.
> Your options are, as I see them, to enlarge your box to avoid this problem,
> use implicit solvent (GB or something), or use restraints to prevent
> rotational motions, though this could be tricky.  You could potentially
> place a couple "ghost atoms" that have no charge, vdw, mass, etc., but whose
> positions you fix via positional restraints.  Then maybe you can constrain a
> dihedral angle between 2 atoms on either side of the dumbbell and the 2
> ghost atoms you fixed.  Or maybe you'd need 2 dihedral restraints... I can't
> quite visualize it right now, but you can use geometry to help here.  In any
> case, it seems like a clumsy hack that's probably not worthwhile anyway (and
> it may have other side-effects that you don't want).
> All the best,
> Jason

AMBER mailing list
Received on Thu Jul 08 2010 - 06:00:03 PDT
Custom Search