[gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center

[XAvier Periole]

My very dear Tsjerk,
> Following this discussion for the past days or so, let's first make
> one thing clear: -pbc nojump is of no use at all when the purpose is
> to keep things together. That only works if everything moves together
> and there is no diffusion, like in protein complexes. For solvent, and
> for bilayers, applying -pbc nojump will actually reveal the diffusion
> across boxes, countering the wrapping that would otherwise occur in
> mdrun (for mathematical/computational reasons). I hope that settles
> that part of the discussion.
Well you can see the bilayer as protein complex where each monomer is
a leaflet. Applying -pbc nojump on the trajectory would simply keep  
the two
leaflets together. Then the system can be centered using the bilayer as
reference: this would put the bilayer at the center of the z axis. The  
movements
"out of the box" can then be corrected using -pbc mol.
It is trivial is the reference structures are correct.
>
> Now the problem here is that the membrane drifts, and the purpose is
> to remove that drift. If it would only be required that the leaflets
> stayed together, that could actually be achieved by removing jumps
> only in the membrane normal direction. So that would call for options
> like -pbc nojump_x |  nojump_y | nojump_z That's actually not that
> hard to do.  Applying such a procedure followed by centering might
> actually come close to what Maria wants to do. The other option is to
> calculate the change in COM from the displacements of particles in the
> periodic system, rather than from the actual COM. Compensating for
> such a shift and then putting all atoms back in the box I think would
> be the solution to the problem. That would call for an option like
> -fit com, that could be used together with pbc options (-pbc nojump |
> inbox | ... ). If required I may be able to implement such changes, or
> provide the routine.
More options like you describe would definitely be very useful.

Mazel
>