[gmx-users] Re: Error- Simulation box resizes during mdrun

[Bharath K. Srikanth]

Hi Justin:

Thanks for your reply. Yes, I did everything you mentioned. Let me explain
my steps in a bit more detail.

1) I had a lipid bilayer and a peptide from a previous simulation (in a
box 15 nm x 7.5 nm x 7.5 nm- 15 nm being perpendicular to the plane of the
bilayer). I removed the water and the peptides, so I had only the bilayer
left.

2) I then resized the box using editconf (the -box option) to a cubical
box of 7.5 nm in all directions. The bilayer remained in the center of the
new box.

3) I then re-inserted the peptide by editing the coordinate file, and ran
an energy minimization. Then I added water using genbox, and ran another
EM.

4) Then I ran the simulation for 3,000,000 steps (90 ns). While this was
happening, I observed that the size of the box was beginning to change, in
the z direction (not in the y direction as I said yesterday, but still,
parallel to the plane of the bilayer).

I've included some pictures here (in the .tga format), from before and
after the simulation. When I ran it in vmd using the initial coordinates
and the trajectory file, I could actually see the box resizing during the
run.

https://www.dropbox.com/l/UJuFbDTPaISIpkSf

I'm sure there's a very simple explanation, but I can't seem to figure it
out.

Thanks!

Regards,
Bharath

On 2/6/13 11:52 AM, Bharath K. Srikanth wrote:
> Hi everyone,
> Today, I was attempting a simulation of a system with a lipid bilayer,
and
> the size of my simulation box, obtained from a previous simulation, was
15
> nm x 7.5 nm x 7.5 nm (15 nm being the direction perpendicular to the
plane
> of the bilayer). The bilayer consisted of 128 DOPC lipids and was
located
> at the centre of the box. I used periodic boundary conditions in all
three
> directions.
> I then resized the box to 7.5 nm in all directions, using the editconf
-box option. After doing the requisite energy minimization steps, I ran
the simulation using mdrun. However, running the simulation caused the
box
> to gradually shrink in the y-direction (parallel to the plane of the
bilayer!) during the simulation, ultimately settling at around half the
original dimension. Is there anything I may have done that could cause
this effect? I'm thoroughly baffled.

What do you mean by "resized the box?"  Did you strip out water, adjust box
vectors, re-solvate, equilibrate, etc?  What you're describing is very weird;
pictures, plots, etc would help a lot here.

-Justin