Re: equilibration of water

From: <>
Date: Mon 29 Mar 1999 08:31:55 -0800

"Marcela Madrid" writes:
>In a system consisting of protein + DNA + counterions +water,
>when I equilibrate the water, do I have to keep protein +DNA + counterions
>fixed, or should I let the counterions move with the water?


I don't know if there is a "truly correct" answer to this question. But,
presuming you started with a vacuum system with a known experimental
configuration (DNA + protein), then you added water and counterions
to it, you should probably start by holding the DNA/protein fairly
rigid using the ntr=1 option, and a fairly high harmonic constant for
the DNA/ protein. I let the counterions move with the water at this
stage because the initial placement of the counterions (at regions of
positive charge) is somewhat arbitrary. During the periodic simulation,
the counterions will form a cloud in the box rather than spending all their
time in the original locations. Eventually the water and counterions
will "equilibrate" as monitored by the average pressure and density.

Once the water box reaches a density approaching 1 , I reduce the harmonic
constant on the dna (or protein) slowly while minimizing the whole system,
then commence production runs with a prmtop that contains the equilibrated
box size of the last dynamics run.

My specific protocol for running periodic systems is as follows.

1) Start with a vacuum solute (DNA/protein/whatever) structure.
Could be experimental, could be generated.
2) Minimize the solute for ~100 steps. This just fixes up any problematic
bond lengths or angles. Don't overminimize as this system is not very
physically realistic (unless you're modelling a solute in the vacuum phase!)
3) Use LEaP or EDIT to add water and counterions. I add the water first,
then the counterions, using LEaP.
4) Minimize the entire system for 1000 steps, holding the DNA rigid
with ntr=1 harmonic restraints.
5) Non-PME dynamics to equilibrate the box (I only use non-PME here because
it's faster). DNA is held rigid with ntr=1 harmonic restraints.
You *could* use belly here but the pressure scaling problem is a pain,
so ntr is preferred unless you're willing to hack the prmtop file.
Counterions and water are free to move at this step- this corresponds to
your question above.
6) optional: some PME dynamics here (I don't think this is necessary-
my feeling is that the first 250-500ps of simulation act as the "PME
equilibration phase"). But I do some PME here just to make sure it works.
7) 5 steps of minimization, slowly lowering the DNA/protein ntr restraints.
You probably want to have periodic boundary conditions enabled during these
minimizations; else a water might move out of the box and cause problems when
you enable PBC later
8) Commence production dynamics. Make sure to use a prmtop which has the
correct box size from steps 5&6 above. Usually 250-500ps are necessary to
get the dna to equilibrate. However, depending on your system, it may
take far longer for the solute to move away from the initial structure.
With your dna/protein system you may find that you have to simulation for
5-10ns for satisfactory results- and even then, that's about 6 orders of
magnitude shorter than these simulations should run.

Received on Mon Mar 29 1999 - 08:31:55 PST
Custom Search