[gmx-users] Fwd: ion channel in lipid bilayer

[alex rayevsky]

I'm sorry for repeat, but nobody answered the question and I decided to
duplicate the request. Maybe it is a problem with a form of question or its
content, please, point me the mistake.

Hi all!

I have a question concerning immersion of the ion channel (four subunits
with extracellular domains and a bundles of helixes)  into the lipid
bilayer. 6 years ago I used some tutorial or mailing lists, which described
the way from KALP15 tutor. With CCR5 model there were no problems at all.
Now I have a not 'cylindric'  protein with a complex shape and overhanging
domains.
the forcefield is CHARMM36, lipid type - POPG.

I tried Membrane builder, but couldn't orient the plane of the membrane by
changing XYZ principles many times in different combinations.. Thus I used
a slightly modified KALP15 method (other .itps, lipids and water
molecules).  First of all after pdb2gmx for protein a series of topologies
were generated with identifiers in the name, as it was assigned in each
chain. However editconf produced a new pdb from the outpu gro without any
ID or terminators for the chains, in Pymol it is represented with a
tetramer entirely highlithing if a single chain is selected (maybe it is a
reason of faults at later stages).

Well, it works fine until the perl script execution. Beside some problems
with the output (system_inflated.gro was corrupted, but I repaired it with
simple python scripting and got a pretty protein in the center of rare
molecules, which looks reliable enough) I started to compact the bilayer to
rich the lipid area of ~53A^2. I finished it on the 13 stage of perl
execution - the distance between nearest lipids was about 16A, however, the
layer was really holey. At the same time the protein was not surrounded
from all sides. But if I try to put lipids closely one to other, they are
simultaneously penetrate the protein body.

Is it the correct method for such kind of simmulations? Could I increase
the number of POPG molecules after getting inflated.gro file with scaled up
bilayer (the initial step for tightning) before scaling iterations? I can
do it manually by copy of the layer (all lipid coordinates) and its
rotation around Y axis in any soft to enlarge the number of molecules in
the cell (even 200 mols is more than 128). Of course I will make changes in
a topology file. It seems, that I will obtain a fully wrapped protein
without anxiety about clashes or presure in cavities...

What do You think? Thank You in advance!