[gmx-users] cutoffs vs pme and lipids

Lynne E. Bilston l.bilston at unsw.edu.au
Thu Jan 23 03:15:53 CET 2003

Dear Erik and Xavier,

Thanks for your thoughts. I'll try increasing the rvdw value for my pme 
simulations, and also try the dispersion correction.  I had let everything 
equilibrate for 1ns in both cases before doing the simulations with the 
anisotropic pressure, so I don't think it's an equilibration issue alone.

What I'm actually trying to do is "stretch" the membrane by applying a 
negative membrane plane pressure (or alternatively, a positive 
perpendicular pressure), to see if I can simulate the effect of membrane 
stretch on my protein.

In the cutoff sims, the lipids stay closer together (yes, smaller area per 
lipid), transmitting more of the pressure to the protein, whereas in the 
pme sims, the lipids space out more, meaning that the protein doesn't see 
the same loading.

Thanks again,

At 12:56 PM 23/01/2003, you wrote:
>From: Erik Lindahl <lindahl at stanford.edu>
>To: gmx-users at gromacs.org
>Subject: Re: [gmx-users] cutoffs vs pme and lipids
>Reply-To: gmx-users at gromacs.org
>Hi Lynne,
>By "separating", do you mean you get a different area per lipid, or?
>In general, lipid systems can be quite pathological :-) Area/lipid is
>extremely sensitive to the balance between attractive coulomb forces and
>entropy in the chain region, so even small changes can have large
>effects (and most forcefields are parameterized using cutoffs for the
>coulomb part).
>There is one significant part of the electrostatics where PME will lead
>to a higher area/lipid: All lipid headgroups have a small component of
>the dipole
>parallel to the membrane normal. All these parallel dipoles will repell
>each other, and with PME we include this effect out to infinity.
>HOWEVER, this
>does not necessarily mean you always get a higer area/lipid with PME
>since there are probably countereffects...
>If you want to compare two simulations it is also a good idea to either
>use the same VdW cutoff distance, or at least apply dispersion corrections.
>Lynne E. Bilston wrote:
> > Hi all,
> >
> > I'm a little surprised at the quite large differences I get in the
> > response of a lipid bilayer when changing from cutoffs to pme. I'm
> > simulating a protein/lipid system (POPC), using Peter Tieleman's popc
> > files, and the usual lipid corrections to the gromacs forcefield.
> >
> > When I use semi-isotropic pressure coupling to apply a pressure to the
> > system, I find that there are big differences in how the lipid
> > responds. The simulations using cutoffs result in the lipid molecules
> > staying together much longer, whereas in the pme simulations, they
> > separate much more quickly, and the box expands much faster. this
> > makes a really big difference in what the protein sees sitting in the
> > membrane. The longer I run the simulation, the more significant the
> > difference (at least up to the 10-12ns I've run on the two systems).
> >
> > For cutoffs, I'm using rlist, rvdw= 1.0, rcoulomb=1.8, and for pme,
> > they're all set to 0.9, based on what  I'd read in the list. All other
> > parameters stay the same between the two simulations. I've looked at a
> > plain lipid system, and see the same thing.
> >
> > Is this big difference an expected thing? Are my mdp settings wrong?
> >
> > Is there a problem with pme and pressure coupling apart from the
> > "artificial ordering in some systems" mentioned in the manual?
> >
> > Any ideas/comments/explanations would be welcome.
> >
> > Thanks,
> >
> > Lynne
>Message: 12
>Date: Wed, 22 Jan 2003 20:55:25 -0500
>Subject: Re: [gmx-users] cutoffs vs pme and lipids
>From: Xavier Periole <periole at inka.mssm.edu>
>To: <gmx-users at gromacs.org>
>Reply-To: gmx-users at gromacs.org
> > This message is in MIME format. Since your mail reader does not understand
>this format, some or all of this message may not be legible.
>Content-type: text/plain; charset="US-ASCII"
>Content-transfer-encoding: 7bit
>Hi Lynne,
>I have experienced the same pattern when I began to
>run simulations of pure lipids. The only thing you
>have to be sure is that your membrane is fully
>equilibrated within the parameters (cutoff, PME,
>pressure coupling, temperature etc ...) you use.
>Specially if you
>use a membrane already equilibrated. My experience
>is that the membrane can take a while to equilibrate.
>No offence to Peter's membrane, they are very useful,
>but if you switch to PME, you better equilibrate it
>again an before putting the protein inside.
>In the simulations I do now I use the semi-isotropic
>control of the pressure, PME (cutoff 0.9nm), and cut
>vdW at 1.2nm. The area/lipid is approximately 64 A*2
>for POPC. Note that when cutting the vdW at 0.9nm I
>had a area/lipid of 68A*2.
>Hope it will help you.
>Physiology & Biophysics Dept.
>Mount Sinai School of Medicine
>New York, NY 10029.
>tel : 1 212 241 8192
>fax : 1 212 860 3369
>e-mail : periole at physbio.mssm.edu
>web-site : http://physbio.mssm.edu/~periole

Lynne E. Bilston, PhD
Senior Scientist, Prince of Wales Medical Research Institute, and
Conjoint Associate Professor, Faculty of Medicine, University of New South 
POWMRI, Barker St, Randwick, NSW 2031 Australia
Tel: +61-2-9382-7924	Fax: +61-2-9382-2643

More information about the gromacs.org_gmx-users mailing list