Re: Membrane Simulations

From: Vlad Cojocaru (
Date: Thu Oct 23 2008 - 02:06:33 CDT

Dear Anshu,

Although I am not that experienced, I believe using NVT over NPT in the
melting phase ensures that the bilayers which are far from equilibrium
do not collapse before being properly solvated in the head-group region
and before the tails acquire their intrinsic disorder. However, its not
clear to me why shouldn't be correct to start directly with NPT
especially if you have membranes which are far from equilibrium such as
those created with VMD. Also, the melting with fixed head groups is done
sometimes even before solvation. If you have resources maybe you should
simply try different protocols. Its always the best solution when you
are not sure. Also, its nice to find such answers by yourself than from
others. If you do not have time and resources, than probably best is to
stick with what is the most documented.

Constant area (or NPgammaT - adding surface tension) is used during
production runs because all currently available all-atom force fields
for lipids do not lead to a nicely equilibrated area per lipid. The
equilibration is very long and it generally leads to an area per lipid
which is smaller than the experimental value. Now, of course keeping the
area constant in the simulation adds stress to the simulation. How this
affects your results its not clear. There is literature on the topic so
better read it before deciding. Its also not clear whether it is really
better to run with an artificially kept constant area or in NPT with
smaller area per lipid than the experimental value. It is common sense
that one should follow more membrane properties than just the area per
lipid to decide.

I guess for a start take a look at:
Kandt et al, Methods 41 (2007) 475488
Gullingsrud J and Schulten K, Biophysical Journal Volume 86 June 2004

and references therein ....

Hope this helps a bit

Anshu Bhatia wrote:
> Hi,
> I asked this question on the forum before but did not get a
> satisfactory response so far. Let me try and rephrase my question.
> I am trying to study the spatial variation in lipid bilayer properties
> due to the presence of a gramicidin channel in the middle of the bilayer.
> Because a gramicidin channel is smaller in length than the lipid
> bilayer thickness, I expect the bilayer to shrink near the channel,
> but retain
> its normal thickness far away from the channel.
> I need help in developing a protocol for my specific problem. From the
> protocols I have seen so far, I have 2 questions for the protocol:
> (1) All protocols seem to constrain the head groups of lipids for a
> short period of the time and let the lipid tails melt before letting
> the whole system evolve.
> This melting is done to interlace the lipid chains without changing
> the thickness. In some protocols this melting is done in NVT ensemble.
> What is the reason to choose NVT over NPT?
> (2) After this melting is performed most protocols use an NPT ensemble
> with constant area or constant ratio for production runs.
> Why is constant area used? In my case, I am interested in looking at
> the variation in properties of lipid across the space, should constant
> ratio still be used.
> P.S.: If anyone can provide me with a good reference paper which
> discusses issues about lipid simulations in namd, I would be very thankful
> Anshu.

Dr. Vlad Cojocaru
EML Research gGmbH
Schloss-Wolfsbrunnenweg 33
69118 Heidelberg
Tel: ++49-6221-533266
Fax: ++49-6221-533298
EML Research gGmbH
Amtgericht Mannheim / HRB 337446
Managing Partner: Dr. h.c. Klaus Tschira
Scientific and Managing Director: Prof. Dr.-Ing. Andreas Reuter

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:48:30 CST