[gmx-users] genbox problem with micelle

[Mark Abraham]

> Dear GROMACS users and developers..
>
> I'm trying to run MD simulation on a micelle structure, which i've built
> using Packmol. The problem is when i use genbox to solvate the system
> using SPC water, the water came inside the micelle, and my simulation
> didn't produce any significant changes to the structure (the behavior of
> lipids which are lipophilic should squeezed the water molecules outside
> the micelle structure). I'm wondering if i can tell genbox not to put
> water molecule inside the micelle...can i?.. :-p

genbox just uses vdw radii for different atom types to generate volumes
that exclude waters. If your micelle (and I presume from the above that
you have a micelle that is lipid-inside-surface water-outside-surface) has
interstices that are large enough to admit waters, then genbox will place
them there.

Probably, an energy minimization of the micelle in vacuo will see it
contract so that the lipids are pretty much close-packed, and hopefully
this density will also be close to that for the micelle in solution.
Alternatively, or if there are still holes that admit waters, you can make
the radii of your lipid C and H atoms larger for the purposes of genbox
using a modified gromacs/share/top/vdwradii.dat. Be careful later with the
gentleness of your minimization and equilibration schemes else you'll
generate large structural changes and thus bad contacts and LINCS errors,
etc.

Otherwise, you could get out a molecule visualization program like VMD,
find the waters inside and delete them (you could script this with VMD,
too).

There's also a lesson here to look at your structure before you simulate
to see that it looks how you expect it to. If you'd seen these waters, and
knew for some reason that it would take longer than your simulation time
to form an equilibrated micelle by excluding these waters, then you'd have
known not to bother with this simulation, and fixed your problem without
spending computer time needlessly.

Mark