[gmx-users] lipid bilayers jumping out of the box

[mark vaughn]

I have been experimenting with making bilayers of complex composition.  
I put together a low-density bilayer from the pdb files of several 
lipids translated to desired points, then rotated to not overlap with 
its neighbors--sort of a home written genconf with the capability of 
using an arbitrary number of pdb structures stacked at arbitrary 
locations. The longest acyl tails of each leaflet are aligned on a given 
z-plane,  then translated so that the initial separation between 
leaflets of 0.2 nm.  I generally center this homemade bilayer on the 
origin. I put it in a vacuum box using editconf with the -d option . VMD 
shows a nice organized bilayer. I expand the z-coord of the box, do an 
energy minimization,then compress it, either with position restrained md 
to a reference configuration or by scaling it to a higher density.

However there is an annoying problem: every time I use mdrun (em, pr, 
whatever) the structure immediately jumps out of the box, so I end up 
with half bilayers with vacuum in between or other weird configurations. 
When i put water in later on, same problem.  Okay, the bilayers can all 
be put back together with trajconv, but it is troublsome to do so.  My 
main question, though, is what is wrong with my original pdb that makes 
the bilayers jump out of the box?

thanks

mv