[gmx-users] Naughty Vacuum Bubble in our Vesicle!

Björn Sommer bjoern at CELLmicrocosmos.org
Fri Dec 19 21:49:15 CET 2014

Dear Andre, Rajat, Stephane, Patrick & Justin, Dear Vesicle Simulators!

Thank your for all your clues, we really appreciate them and they helped 
us to solve our problem!

Some weeks ago, I asked you about my problem while simulating a vesicle 
using united-atoms method (ffG45a3).

The main problem was the forming of a vacuum in the internal side of the 
vescile when I started the
equillibration after removing water out of the membrane. The vacuum 
bubble was created because the lipid packing was too dense, which 
resulted in a relaxation of the vesicle and a volume enlargement. I 
solved this problem and just wanted to tell you how I did this.

First of all I removed, as you suggested, less water, by using my TCL 
script. The vaccum bubble was still created but I just used genbox a 
second time and equillibrated again, after this, no vacuum was found. 
(The repetition of the genbox might have to be repeated a second time, 
depending on the packing density.) The production run gave me no 
problems at all and my simulation worked out pretty well. I used some 
self-written python scripts to analyse the vesicle and found out it was 
relatively stable. Also, there is the TCL script which I used to remove 
the water from the vesicle membrane.

If you're interested, you can find these scripts here:


(Please copy this link to your browser - I had some problems sending the 
email with the full URL to the user list.)

We are happy about critics or suggestions for improvement of the scripts!

Best wishes,
Manuel and Björn

P.S.: Some additional comments about your suggestions: we wanted to use 
a direct method, using the membrane model from the MembraneEditor and 
then just starting a MD simulation with ffG45a3. The idea with the pore 
and the MARTINI force field is very good and interesting, but would be 
too complicated for our purposes. Because time was limited for this 
project, we also skipped the (very interesting) idea with the surface 
tension computation, because the method and its implementation seems to 
be bit time-consuming.

More information about the gromacs.org_gmx-users mailing list