Re: Simulation issues: water box deformation after minimization and shifting protein coordinates during production run

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Mon Nov 10 2008 - 13:52:11 CST

Mamta Mohan wrote:
> Dear Peter,
>
> Perhaps I missed some explanation in the previous mail.
>
> The periodic cell I defined I used minmax and listed the coordinates
> (I can list the macro if you want). The coordinates of the box are not
> touched once box was built. The box itself seems to have shifted after
> minimization. In the snapshot I redraw to box to explain my point.
My point is that measuring the minimum and maximum coordinates in vmd is
meaningless if some components of the system are not wrapped, and that
the reason that your water box appears to "jump" is that you *are*
wrapping the water coordinates into a box with a different center from
the start of the simulation (unless I misread your original email, which
seems to indicate that your pre-simulation box is centered near
(0,0,0)). Rather than measuring the system minmax in vmd to figure out
how big the actual periodic box is after the simulation, look at the box
size and center in the .xst file from namd. The box you've drawn in your
snapshots is meaningless because you're not measuring it based on what
the actual periodic cell is.

Try drawing the box present in your .xst file before, during, and after
the simulation.
>
> Minimization and production ran as one step (please take a look at
> config file)
>
> The snapshots I took by superimposing the two molecules (as explained
> below in the mail) to make my point. First molecule is before
> minimization and production run, that is psf+pdb and second is after
> simulation finished psf+dcd.
>
>> From what I see, water box shifted and protein seems to be drifting
> constantly towards the top of the box during production run.
>
> I also think (and please explain me why if I am incorrect) that I do
> not have to wrap proteins and ions, I think wrap water should be OK. I
> can wrap dcd later for protein and ions.
You don't have to; it is fine to wrap whichever system components you
want unless you're doing something like a diffusion constant
calculation. However, you cannot then trust the results of measure
minmax to tell you what your periodic cell size is, because some of the
atoms are not *in* the coordinates of the unit cell (indeed, you never
can if you have a protein, since its coordinates will not always be
contained within the periodic cell).

Peter

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:48:36 CST