[gmx-users] Actual box size

[Justin Lemkul]

On 12/17/12 1:48 PM, zugunder wrote:
> Hi,
>
> I want to simulate a water soluble protein of approx. 160aa size and its
> shape from a .pdb looks a little elongated (if rotated around the longest
> axis, it roughly occupies a cylinder with length to diameter ratio 1.8 -
> 1.9). Would it be considered globular enough to try dodecahedron box?
>

In almost all cases, a dodecahedral box is the optimal choice.  A cubic box with 
the same periodic distance for an elongated protein would be much larger.

> Out of curiosity, I compared the triclinic and dodecahedral boxes created
> for my protein.
>
> After I created a triclinic box with g_editconf, I got this:
>
> Read 2406 atoms
> Volume: 83.8247 nm^3, corresponds to roughly 37700 electrons
> No velocities found
>      system size :  4.830  3.541  4.901 (nm)
>      center      : 19.439-14.327 13.948 (nm)
>      box vectors :  4.830  3.541  4.901 (nm)
>      box angles  :  90.00  90.00  90.00 (degrees)
>      box volume  :  83.82               (nm^3)
>      shift       :-16.024 17.097-10.498 (nm)
> new center      :  3.415  2.771  3.451 (nm)
> new box vectors :  6.830  5.541  6.901 (nm)
> new box angles  :  90.00  90.00  90.00 (degrees)
> new box volume  : 261.17               (nm^3)
>
> Dodecahedron:
>
> Volume: 83.8247 nm^3, corresponds to roughly 37700 electrons
> No velocities found
>      system size :  4.830  3.541  4.901 (nm)
>      diameter    :  5.907               (nm)
>      center      : 19.439-14.327 13.948 (nm)
>      box vectors :  4.830  3.541  4.901 (nm)
>      box angles  :  90.00  90.00  90.00 (degrees)
>      box volume  :  83.82               (nm^3)
>      shift       :-13.508 20.257-11.153 (nm)
> new center      :  5.931  5.931  2.796 (nm)
> new box vectors :  7.907  7.907  7.907 (nm)
> new box angles  :  60.00  60.00  90.00 (degrees)
> new box volume  : 349.61               (nm^3)
>

In the absence of the actual editconf commands, it's hard to judge the utility 
of each of these setups.

> So from the box volumes I see that in this case a triclinic box is a
> preferred choice. However, as I am aware that dodecahedral box
> representation (as seen in VMD for example) is subject to periodicity
> effects, I am not completely sure if exactly these box sizes will be used in
> simulations. If not, how could I estimate the actual box volumes before I
> start simulation (to maximize performance)?
>

The box vectors shown are the ones that will be used unless you manipulate them 
in some way.

> And I have one more short question:
> Is it possible somehow to visualize the "actual" dodecahedral box at this
> stage (i.e. right after adding water, having no trajectory file yet)?
>

You need a .tpr file and can re-wrap the periodic image with trjconv -pbc mol 
-ur compact.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================