[gmx-users] mdp options for charmm27 force field in gromacs version 4.6.7

Tue May 19 14:36:46 CEST 2015

Hi people,

On Tue, May 19, 2015 at 11:38 AM Piggot T. <T.Piggot at soton.ac.uk> wrote:

> Hi Chris,
>
> As far as I understand it, before GROMACS 5 the potential rather than the
> force was switched off. Since GROMACS 5 you can choose between the two.
>

No, as Chris originally asked, and Justin has now also said, there's been a
force switch (= potential shift) since living memory, e.g. compare manual
sections 4.1.5 at least as far back as 4.5.6. The documentation and
people's understanding hasn't been great (and more recent docs have tried
to improve this). I plan completing the clean-up operation when the group
scheme is removed (after 5.1) now that maintaining backward-compatible
execution paths is no longer a concern, particularly for simulation
protocols we no longer think are a good idea. You can see a sketch of the
future in the vdw .mdp options section of the manual for 5.0.5

One noteworthy point is that in 4.6.x, the Verlet scheme did not yet
support force switch, so such a simulation could not run on e.g. GPUs. This
was added in 5.0.

AFAICR rvdw-switch has no effect in the Verlet scheme if vdw = cutoff in
4.6.x, because the available modifiers in 4.6.x do not include any
switches. As I imply at http://redmine.gromacs.org/issues/1413, I don't
think there's a decent way to implement CHARMM non-bonded settings in the
Verlet scheme in GROMACS 4.6.x. Whether some approximation works well
enough is your call.

As for the settings to use with the CHARMM force field, the CHARMM package
> itself (as Justin has said) uses force switching. However, in NAMD vdw
> force switching has only implemented from versions 2.7b3 onwards. For
> proteins, I cannot see this choice having a large impact as the paper
> implementing the CHARMM force field in GROMACS (Bjelkmar et al.) will have
> used the potential switch and all the

That paper used GROMACS pre-4.5, so definitely the group scheme. Poke the
coresponding author (Erik) if you want to find out which vdwtype was used;
the text is ambiguous.

Mark

simulation in NAMD prior to version 2.7b3 will have also used this
> approach. The one case in which I saw this choice having a large impact was
> for CHARMM36 membranes and when a combination of normal TIP3P and potential
> switch was used. When combining these two, this resulted in DPPC membranes
> transitioning to a gel phase at 323K. If either of these was changed (i.e.
> if either CHARMM TIP3P/TIPS3P was used or the forces were switched off) the
> membranes behaved well.
>
> So I would say that if you are using GROMACS prior to version 5 it should
> be fine (with the CHARMM TIP3P water, especially if using CHARMM36 DPPC
> membranes) but it is worth bearing in mind that this is slightly different
> to how it is done in the CHARMM package. Once again, this is just my
> understanding of the way it has worked in GROMACS versions. One of the
> developers who implemented the switch cut-offs (Berk?) should be able to
> confirm (or deny!) this.
>
> Cheers
>
> Tom
>
> ________________________________________
> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se [
> gromacs.org_gmx-users-bounces at maillist.sys.kth.se] on behalf of
> Christopher Neale [chris.neale at alum.utoronto.ca]
> Sent: 19 May 2015 04:20
> To: gmx-users at gromacs.org
> Subject: Re: [gmx-users] mdp options for charmm27 force field in gromacs
> version 4.6.7
>
> Dear Justin:
>
> Thank you for the suggestion. I don't use gromacs 5 because of things like
> this: http://redmine.gromacs.org/issues/1603 that tend to pop up early in
> a release series. Until I needed to run charmm simulations, I use 4.6.7
> because it works and I am confident in it. But your suggestion is really
> good... use gmx 5 to get energies to test against gmx 4 and see if I can
> find valid settings.
>
> Thank you,
> Chris.
>
> ________________________________________
> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <
> gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Justin
> Lemkul <jalemkul at vt.edu>
> Sent: 18 May 2015 20:29
> To: gmx-users at gromacs.org
> Subject: Re: [gmx-users] mdp options for charmm27 force field in gromacs
> version 4.6.7
>
> On 5/18/15 2:21 PM, Christopher Neale wrote:
> > Dear Users:
> >
> > I would like to use the charmm27 force field in gromacs version 4.6.7
> and I would like to know if it is possible to get the proper treatment of
> vdw interactions.
> >
> > Information for gromacs 5 is here:
> > http://www.gromacs.org/Documentation/Terminology/Force_Fields/CHARMM
> >
> > However, Gromacs 4.6.7 does not contain the option for vdw-modifier =
> force-switch
> >
> >    Invalid enum 'force-switch' for variable vdw-modifier, using
> >    'Potential-shift-Verlet'
> >    Next time use one of: 'Potential-shift-Verlet' 'Potential-shift'
> 'None'
> >    'Potential-switch' 'Exact-cutoff'
> >
> > Is a potential-shift the same as a force-switch and therefore the
> following options are correct for charmm27 in gromacs 4.6.7?
> >
> > ns_type = grid
> > cutoff-scheme = verlet
> > vdwtype = cut-off
> > rlist = 1.2
> > rvdw = 1.2
> > rvdw-switch = 1.0
> > rcoulomb = 1.2
> > vdw-modifier = Potential-shift-Verlet
> > coulombtype = PME
> >
> > I know this has been discussed before on the list, but I can't find
> anything that seems authoritative for gromacs versions prior to v5.
> >
>
> I cannot personally attest to anything aside from what's on the wiki.  We
> have
> verified the forces are equivalent between CHARMM's vfswitch and GROMACS'
> force-switch modifier.  This is how we do all our simulations, so for
> posterity,
> what I wrote on the wiki is what I must emphasize is known to be correct.
>
>  From the manual, it seems that the group scheme + vdwtype = shift is
> equivalent
> to the modern force-switch modifier.  The language of potential-shift does
> not
> sound as if it is the same as force-switch, and I don't know why such
> redundancy
> would be present anyway; perhaps legacy naming reasons.
>
> Any reason why you can't use 5.0 with the known correct settings?  The
> check I
> would do is a single-point energy with the settings you think are
> equivalent and
> want to test and print out the forces with gmx traj.
>
> -Justin
>
> --
> ==================================================
>
> Justin A. Lemkul, Ph.D.
> Ruth L. Kirschstein NRSA Postdoctoral Fellow
>
> Department of Pharmaceutical Sciences
> School of Pharmacy
> Health Sciences Facility II, Room 629
> University of Maryland, Baltimore
> 20 Penn St.
> Baltimore, MD 21201
>
> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
> http://mackerell.umaryland.edu/~jalemkul
>
> ==================================================
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>