membrane bilayer crosses the z axis boundaries of the box and separates into two layers

From: Thomas Evangelidis (
Date: Wed Dec 11 2013 - 14:29:33 CST

Dear NAMD community,

I was running a protein-membrane accelerated MD simulation in a hexagonal
prism unit cell and after some time I took the final snapshot, protonated a
residue and recreated the AMBER .prmtop and .inpcrd files. Then I launched
a new accelerated MD simulation by using the cell basis vectors and origin
written in the last .xsc file of the previous simulation. These are the
relevant parts of my configuration file:

firsttimestep 0
temperature $temperature
cellBasisVector1 93.268 0 0
cellBasisVector2 46.6341 80.7725 0
cellBasisVector3 0 0 94.3582
cellOrigin 60.4405 59.5913 43.92

wrapWater on
wrapAll on
wrapNearest on

PMEGridSpacing 1.0

useGroupPressure yes
useFlexibleCell yes
useConstantRatio yes
SurfaceTensionTarget 26

reinitvels $temperature
run 6000000

I used constant surface tension as required for the lipid11 AMBER ff. My
problem is that after a few nanoseconds the membrane bilayer crosses the
plane perpendicular to the z-axis and the two lipid layers separate. It
seems that the barostat is acting on the lipid tails (???) and creates an
empty space between them that spans the other two periodic cells on the
z-axis. I have uploaded a "before" and an "after" pic (the atoms of the
central box are rendered in CPK mode):

It is worth noting that the membrane bilayer was not at the center of the
box in the beginning and the cellOrigin was poised below it. That was a
consequence of the continuation from a previous simulation.

How can I prevent the membrane bilayer separation? Can I set the cellOrigin
to be at the center of the bilayer?

I would be very grateful for any advice.


Thomas Evangelidis
PhD student
University of Athens
Faculty of Pharmacy
Department of Pharmaceutical Chemistry
157 71 Athens


This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:59 CST