Re: use of charmm27 & NPT ensemble for lipid bilayer simulation

From: JC Gumbart (
Date: Sat Jul 12 2008 - 22:07:00 CDT

Your specific issue, with lipid tails overlapping, is a good reason
to equilibrate the tails first by restraining everything else in your
system (including the headgroups) for maybe 0.25-0.5 ns. This will
"melt" the tails, preventing the bilayer from shrinking quickly
because of the initially straight tails.

Regarding your more general question, the answer is not so
straightforward. One cannot assume that area per lipid is the only
parameter worth optimizing, or that if one gets it right, everything
else falls in line. I've seen cases where a bilayer constrained to a
specific area/lipid will display other strange behavior such as
freezing in some places. So generally, I will not apply constant
area unless for a specific reason or that I feel that the bilayer
shrinks TOO much (some will shrink drastically). If the change in
area is not that much during equilibration and then stabilizes, I
don't concern myself with how accurate the area/lipid is and instead
focus on getting reasonable behavior in general from the lipids.

These considerations are usually sufficient since for many cases, one
just wants to roughly approximate a bilayer anyway. However, if your
protein is mechanosensitive (and many more proteins than we realize
may be!), you must obviously pay much closer attention to the
pressure profile of your bilayer.

On Jul 11, 2008, at 12:26 AM, Ruchi Sachdeva wrote:

> Dear all,
> I am new to lipid bialyer simulations and facing the following
> problem. I have equilibrated my system containing a membrane
> protein embedded in POPC bilayer using charmm27 & NPT ensemble on a
> linux cluster. I found that lipid tails are overlapping with each
> other & thus bilayer thickness is getting reduced throughout the
> simulation.
> Then I found a few namd posts (link1, link2) mentioning that use of
> charmm27 & NPT does not give correct lipid bilayer parameters such
> as area per lipid (by Jensen et al., 2004) The parameters deviate
> from the experimental values. Instead one should initially
> equilibrate holding the lipid head groups z coordinates in NPT
> ensemble. Once area per lipid is set, then use constant area ie.
> NPzAT ensemble. Or one can use NPgammaT simulation also.
> Despite this report by Jensen et al., I found papers still using
> charmm27 force field alongwith NPT ensemble, which is not
> recommended. Moreover, nothing is mentioned about lipid parameters.
> Can anyone please shed some light on this. Am I missing something?
> Thanks and Regards
> Ruchi

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:39 CST