Tutorial-l Mailing List

From: Timothy Isgro (timisgro_at_ks.uiuc.edu)
Date: Fri Nov 03 2006 - 14:33:18 CST

Hi,

I think the most viable explanation of the phenomenon of drifting
components in a periodic system is some degree of insufficient
"equilibration".

The problem should not lie in explanations 2 and 3 that you suggest,
whereby the periodic images affect the movement of the bilayer in the
z-direction. If you try to reason that a movement downward in the z-
direction is caused by the periodic image below "pulling the system
down" (or the image above repulsing it), simply imagine an infinite
array of z-periodic images for yourself, and you can see while the
bilayers may change position relative to a fixed grid (the periodic
box), they do not change position relative to one another. By
Newton's 3rd law, the "up" forces between periodic bilayers will be
countered by the "down" forces. The two are identical.

As I mentioned, I think that some form of insufficient equilibration
could be responsible for movement within the periodic box. I explain
using the following analogy. Imagine a molecular system (protein)
shaped like a bird. The unequilibrated form of this system has two
"wings" extended from the protein like a gliding bird would, and the
equilibrated system has the wings contracted toward the body as if
the bird has flapped the wings to provide thrust. Placing the
unequilibrated protein in water and allowing unrestrained dynamics
would result in an equilibration whereby the protein "wings" snap
toward the body, propelling water leftward, say, in the periodic box,
and thrusting the protein rightward.

I hope the analogy makes sense. It is a rather drastic example
designed to show how an unequilibrated protein might move in the box,
and I think it is important to realize that less drastic
"unequilibrations", such as small rearrangements in lipid positions
for instance, can result in similar movements. In this respect, it
is very important how one defines equilibration relative to the
properties of the system being probed. Dynamic systems may be in
equilibrium with respect to one property or scale, but not another.
A small rearrangement of lipids may be considered inconsequential,
but a large change in protein structure would not be. However, both
may result in the system drift you see.

-Tim Isgro

On Nov 3, 2006, at 6:57 AM, DL Lynch wrote:

> Hi,
>
> I ran the tutorial:
> http://www.ks.uiuc.edu/Research/namd/tutorial/NCSA2002/hands-on/
>
> I built the system as described in the tutorial.
> I minimized and equilibrated the systems as described in the tutorial
> (equil.namd from the above website).
>
> Then I ran the nptsim.namd. but instead of running for
> run 20000 as described in the tutorial, i ran for 2ns (2000000).
>
> When I did this the system seems to drift in the z direction. In
> the start
> of the simulation the water's are equally spaced above and below
> the bilayer,
> however at the end of the 2ns (actually it doesn't take too long to
> see the
> following effect, but I grabbed the last image) in 2ns_of_sim.png
> you can
> see that the bilayer has "drifted" lower in the simulation cell. (P
> atoms are the gold balls)
>
> http://science.kennesaw.edu/~dlynch/2ns_of_sim.png
>
> Do you have any thoughts as to what could be causing the
> center of mass of the water to go "up" and the bilayer to go "down"?
>
> Is this an issue of:
> 1) poor equilibation from the equil.namd step? Meaning:
> a) the run where the protein is fixed is not long enough?
> b) and/or the run where it is constrained is not long enough?
> c) and/or the NPT with large damping coefs. is not long
> enough?
> d) and/or the damping coefs are made smaller (in the
> nptsim.namd file) in too bigh a step?
> 2) the POPE are "seeing" each other in the periodic images?
> 3) the peptide is "seeing" itself in the other images?
>
> The reason I am trying to sort this out is that I am seeing a
> similar behavior in
> my simulations and so I tried the tutorial to see if I could sort
> out the problem
> on a smaller system.
>
> I am going thru 1a-1d but I thought someone may have addressed this
> already.
>
> Any thoughts or suggestions on this matter would be greatly
> appreciated.
>
> Thanks!
>
> ps This question has been touched on in the NAMD mailing list, but
> I was not able to see
> if it was resolved.
>

**
** Visit http://www.greatcircle.com/majordomo/ for more info on majordomo
**