[gmx-users] CHARMM36 lipid bilayers

Sven Jakobtorweihen Jakobtorweihen at tuhh.de
Thu Oct 21 15:57:27 CEST 2010


Hi Berk,

the original charmm article used
rvdw=1.2
rvdw_switch=0.8
coulombtype= PME
Unfortunately, they did not specify rcoulomb explicitly. In  the
discussions they  wrote ".. simulations of bilayers with C36 should be
carried out with PME  with rc=10 or 12 A and no long-range corrections
for the LJ term." I tested also  with rcoulomb=1.2, but the results were
closer to Klauda with rcoulomb=1.0. However, I think by adjusting
fourierspacing and pme_order I probably could get the same results with
rcoulomb=1.2 as with rcoulomb=1.0.

rlistlong=1.4 I set to avoid the note that it should be 0.1 to 0.3 nm
larger than rvdw. I realized that g_tune always sets rlistlong to the
largest cutoff and, hence, ignores the note. Considering efficiency I
would from now on use rlistlong=1.3 and if someone could convience me to
ignore the note I would go with 1.2. By the way, when g_tune changes
rlistlong in the tpr files where the load is shifted from the reciprocal
to the real space part the most (if not all) performance gain compared
to the original tpr comes from the rlistlong change.

See you,
Sven


Berk Hess schrieb:
> Hi,
>
> You have very strange and complex cut-off settings in Gromacs.
> What Charmm settings are you trying to mimic?
>
> Berk
>
> > Date: Thu, 21 Oct 2010 15:03:51 +0200
> > From: Jakobtorweihen at tuhh.de
> > To: gmx-users at gromacs.org
> > Subject: [gmx-users] CHARMM36 lipid bilayers
> >
> > Dear gmx-users,
> >
> > recently Pär Bjelkmar and Thomas Piggot have generated force field files
> > for Charmm36 lipids. I run some simulations to find the best run
> > parameters and to check if the results of the original Charmm36 lipid
> > article [Klauda et al., J. Phys Chem. B, 2010, 114, 7830) can be
> > reproduced with gromacs.
> >
> > I run 40 ns NPT simulations with semiisotropic pressure coupling
> > (Parrinello-Rahman, tau_p=5), the first 10 ns are equilibration and
> > averages were calculated for the last 30 ns. DMPC and POPC at 303 K and
> > DPPC at 323.15 K (Nose-Hoover, tau-t= 1). The itp files were made with
> > pdb2gmx -nochargegrp. All simulations contained 128 lipids and
> > approximately the same water/lipid ratio (water is TIP3P) as Klauda et
> > al. I started from charmm27 bilayers provided at the Chramm Gui website.
> > I used the following parameters:
> >
> > rvdw=1.20; rvdw_switch=0.80; DispCorr=No; coulombtype= PME;
> > rcoulomb=1.00; fourierspacing=0.15; pme_order=6; rcoulomb_switch=0.00;
> > nstlist=10; rlist=1.00; rlistlong=1.40; constraints= hbonds; dt= 0.002
> >
> > These simulations result in the following area per lipid [A^2/lipid]:
> > DMPC=56.6 +/- 0.4 ; POPC =61.8 +/- 0.4 ; DPPC=55.0 +/- 0.7
> >
> > Comparing to the results of Klauda et al (all simulation with the
> > charmm-package, except one):
> > DMPC=60.8 +/- 0.2 ; POPC=64.7 +/- 0.2 ; DPPC=62.9 +/- 0.3 ; DPPC=59.1
> > +/- 0.4 (with NAMD)
> >
> > It is obvious that my simulations with gromacs 4.5.1 give lower areas
> > per lipid for all cases. Considering the deviations observed by Klauda
> > et al. between Charmm and NAMD simulations ( rvdw_switch was only
> > changed slightly in NAMD) could lead to the conclusion that DMPC and
> > POPC are fine. But I am a bit worried about the DPPC result. Did anyone
> > have suggestions how to improve it? Are these differences expected when
> > comparing gromacs and charmm simulations? Did by any chance someone else
> > tested charmm36 bilayers in gromacs?
> >
> > Thanks,
> > Sven
> > --
> > gmx-users mailing list gmx-users at gromacs.org
> > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> > Can't post? Read http://www.gromacs.org/Support/Mailing_Lists



More information about the gromacs.org_gmx-users mailing list