AMBER: vacuum bubbles in system equilibration

From: Marie-Pierre Durrieu <>
Date: Tue, 01 Feb 2005 11:56:21 +0100

Dear all,

During setup of a system solvated in water, I observed a problem with
the size of the box created by Leap, as a consequence of which the water
behaves in a strange way leading to vacuum "bubbles". Similar things
seem to happen for some of the tutorials that come with Amber.

I have created in Leap (AMBER 8) a rectangular box of water around my
protein, then I have minimized and heated the system from 100K to 300K
with constant volume. During the heating, huge pockets of vacuum appear
in the corners of the rectangular box in the reimaged trajectory.
Finally the system looks like "a bottle".

Of course the size of the box doesn't change during the heating with
constant volume. If I visualize the shape of the box in VMD, it stays
as a perfect rectangle, and its dimensions are the ones given previously
by Leap. The shape of the water in this rectangle is like a "bottle".
(see <>)

Wondering about so much vacuum, I've visualized the system before the
minimization, and surprisingly the box created by Leap is slightly (~3
A) bigger than the rectangular system of {water+protein}. Layers of
vacuum are present on each face of the periodic system. (see
I first thought that it was caused by the VDW definition of the box
dimensions in Leap : indeed the heigth of the vacuum layers corresponds
to twice the radius of water. But when I continue the equilibration by
an MD run with constant pressure, the box dimensions shrink, and the
shape of the box matches perfectly the rectangular system, with no
layers of vacuum, sweeping off my VDW explanation (see

I've read from the mailing list that "LEaP puts a small 'gap' between
boxes so that there are no bad atom-atom contacts in the initial
structure". In my case the distance between 2 boxes is 3A. Isn't it too
big for a small gap ?
The AMBER 4.1 bugfix 31 has a similar problem to solve : the problem
was "layers of vacuum appear when solvating with WATBOX216" in Leap, and
the cause was "A box including VDW radii is built around solvent units -
this is misapplied when the solvent is already an equilibrated box with
interpenetrating VDW at the box boundary". Do I experience the same
problem with my system ?

I've tried the same minimization+heating with the size of the "bounding
box for atom centers" (given by Leap) inserted in the prmtop, instead of
the "vdw box". During the heating vacuum appears again in corners (see , but it is less
spectacular as for the "vdw box".

Wondering if this problem of box size was characteristic for my system,
I've visualized the reimaged trajectory in the polyAT tutorial. During
the heating from 0 to 300K, vacuum appears on the faces of the
octahedron, deforming the system. On one face the DNA seems to be in
contact with vacuum instead of water. (see

Despite the fact that the MD with constant pressure quickly removes the
vacuum, is it a good thing to start with a "vdw box" that seems to
contain too much vacuum ? Wouldn't it be better to start with a
"bounding box for atom centers" ?
When solvating with Leap, are the small boxes of water already
equilibrated ? If they are, why is there so much vacuum ? Or is this a
problem of the solvent box used to solvate the system ?

Thanks in advance for your help,


Marie-Pierre Durrieu
Institut de Biologie Physico-Chimique
13 rue Pierre et Marie Curie
75 005 Paris - FRANCE
e-mail :
Tel: +33 15841 5161
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Tue Feb 01 2005 - 11:53:00 PST
Custom Search