Date: Thu, 29 Apr 1999 10:40:55 -0700 (PDT)
From: Thomas Cheatham
Subject: protein is compressed upon equilibration

> is a non zero rmsd with the starting structure. It seems that the
> protein is being compressed. I obtained similar results when
> constraining the protein using a harmonic potential.

With constant pressure applied, the restrained coordinates or those held
fixed with the belly option are not truly fixed since the coordinates are
still scaled up and down as the box size changes to maintain constant
pressure.

Although you could switch to NVT, i.e. a constant volume simulation, given
that your protein was shrinking suggests that the box was a little too big
at the start, and therefore during equilibration at constant volume vacuum
"bubbles" may appear as the solvent equilibrates. These are best avoided.

To avoid the scaling problem of the fixed/restrained protein, you can
switch the constant pressure to scale by molecule rather than by atom by
setting NPSCAL=1. If you only have a single protein molecule (and it is
all contained within a single molecule which is the default behavior),
this will allow constant pressure with the belly or restraint to work
as you expect. If on the other hand you have more than one molecule fixed
(such as with a protein protein complex or with a DNA double strand) there
is no general way around the problem without "tricking" sander by
modifying the prmtop; this is discussed in more detail in the AMBER
polyA-polyT tutorial at

http://www.amber.ucsf.edu/amber/tutorial/polyA-polyT/equilibration.html

Good luck with your simulations.

Thomas Cheatham, III