Re: [AMBER] CHARMM Forcefield in Amber - which barostat and ntp to use?

From: Sally Pias <sallypias.gmail.com>
Date: Thu, 4 Jun 2020 21:16:59 -0600

Hi Amanda,

I think the barostat=2, ntp=2 combination should be fine, especially if you
have a membrane protein in there. First, that can make the subtle physical
properties of the lipids less of a dominating factor, as the protein plays
a significant role in the membrane structure. Second, in our experience,
the presence of a protein helps to avoid problematic shape change in long
simulations where x and y are not coupled - as is the case with the Amber
implementation of anisotropic coupling. If it is helpful, you can reference
our paper below as an example of work that uses barostat=2 with ntp=2.

Dotson RJ, Smith CR, Bueche K, Angles G & Pias SC (2017). Influence of
cholesterol on the oxygen permeability of membranes: Insight from atomistic
simulations. *Biophys J* *112,* 2336–2347.

I suppose there is a possibility - as I think one other response mentioned
- that your bilayer might be rotated with the respect to the reference axis
for the anisotropic coupling. Namely, the barostat may be attempting to
couple x and z, for example. An Amber developer might know whether this
makes sense, given the way the coupling algorithm works. I could check
whether the orientation of your starting structure is typical of systems we
have simulated, if you sent me your starting pdb file (with your advisor's
permission).

Best,
Sally

Sally Pias
New Mexico Tech

On Thu, Jun 4, 2020 at 8:46 PM Amanda Buyan <Amanda.Buyan.anu.edu.au> wrote:

> Dear Sally,
>
> I'm using a square bilayer as built off of CHARMM-GUI. I'm more concerned
> with the membrane protein than the lipids at the moment, but would you
> recommend using ntp=2 with barostat=2 for these simulations? I've heard
> different recommendations, but there doesn't seem to be a general
> consensus....
>
> Cheers,
> Amanda
>
>
> Dr. Amanda Buyan
>
> Postdoctoral Fellow, Corry Lab
>
> Research School of Biology
>
> 134 Linnaeus Way
>
> Canberra, ACT 2601
>
> amanda.buyan.anu.edu.au
>
> ________________________________
> From: Sally Pias <sallypias.gmail.com>
> Sent: 05 June 2020 07:08
> To: AMBER Mailing List <amber.ambermd.org>
> Subject: Re: [AMBER] CHARMM Forcefield in Amber - which barostat and ntp
> to use?
>
> Dear Amanda,
>
> Can you clarify the shape of the bilayer you are using as a starting
> structure? As far as I know, Amber only supports square or rectangular
> bilayers. A system snapshot, before and after running in Amber, would be
> helpful. You might send it to me individually, if not to the list.
>
> I do not anticipate a problem with using ntp=3 (semiisotropic scaling)
> with barostat=2 (MC barostat), but this combination has not - to my
> knowledge - been rigorously tested. My group has tried it, with csurften=3
> (constant surface tension in xy plane) and gamma_ten=0.0 dyne/cm. This
> effectively couples the x and y dimensions, preventing box shape change in
> long simulations that can be problematic, especially using the MC barostat.
> We seem to get more or less the same physical behavior using this approach,
> compared with ntp=2 (anisotropic scaling), which has also not been tested
> rigorously with bilayers, to my knowledge. However, applying a surface
> tension of zero may not be entirely rigorous scientifically. Though the
> surface tension of perfectly symmetric bilayers "should" be zero, there are
> likely fluctuations and transient asymmetries that may be important for
> some research problems.
>
> Note that Lipid14 - and possibly also later Amber lipid force field
> versions - were validated and tested with the Berendsen barostat
> (barostat=1). I do not know what barostat was used for Charmm36m. The MC
> barostat does, in our experience, change the bilayer properties of Lipid14
> POPC somewhat, with a tendency to reduce the area per lipid and increase
> the thickness (unpublished results). We find these variations to be within
> experimental error for area per lipid but possibly a bit high (by ~1
> angstrom) for bilayer thickness. My group does routinely use the MC
> barostat with anisotropic scaling for bilayer simulations, but there are
> definitely accuracy considerations that you would want to monitor. I
> believe Rich Pastor at NHLBI has done some recent testing of a Charmm lipid
> force field with the MC barostat, but I do not think the work has been
> published yet.
>
> Best,
> Sally
>
> Sally Pias
> New Mexico Tech
> sallypias.gmail.com
>
> On Wed, Jun 3, 2020 at 6:31 PM Amanda Buyan <Amanda.Buyan.anu.edu.au>
> wrote:
>
> > Dear Matias,
> >
> > Not a silly question at all! I have just checked it, and there doesn't
> > appear to be any gaps or spaces between the lipids in the periodic
> images,
> > so I don't know if "excessive" space is an issue, but it is helpful
> > eliminating one possibility!
> >
> > Cheers,
> > Amanda
> >
> > ________________________________
> > From: Matias Machado <mmachado.pasteur.edu.uy>
> > Sent: 04 June 2020 00:51
> > To: AMBER Mailing List <amber.ambermd.org>
> > Subject: Re: [AMBER] CHARMM Forcefield in Amber - which barostat and ntp
> > to use?
> >
> > Dear Amanda,
> >
> > A very silly question, have you checked the PBC condition before running
> > the simulation? I mean, have you visualized the periodic images of the
> > membrane before any simulation?
> >
> > I don't know how CHARMM-GUI prepares the system for running in AMBER
> code,
> > but there are some issues when setting the periodic box in LEAP for
> > membranes which leads to "excessive" space between lipids of neighboring
> > images breaking the continuity of the membrane. That kind of issue may
> > create odd behaviors at edges...
> >
> > Hope it helps...
> >
> > Matias Machado
> >
> > ------------------------------------
> > PhD.
> > Researcher at Biomolecular Simulations Lab.
> > Institut Pasteur de Montevideo | Uruguay
> > [http://pasteur.uy/en/labs/biomolecular-simulations-laboratory]
> > [http://www.sirahff.com]
> >
> > ----- Mensaje original -----
> > De: "Batuhan Kav" <bkavlist.gmail.com>
> > Para: "AMBER Mailing List" <amber.ambermd.org>
> > Enviados: Miércoles, 3 de Junio 2020 10:44:06
> > Asunto: Re: [AMBER] CHARMM Forcefield in Amber - which barostat and ntp
> to
> > use?
> >
> > Hello,
> >
> > I don’t have access to the data anymore but I remember that MC barostat,
> > at least in case of lipid bilayers, had issues with properly
> equilibriating
> > the structure. If I recall correctly, for instance, that the
> xy-dimensions
> > of the box didn’t change considerably from the initial values and the
> area
> > per lipid values did not converge. I think using Berendsen barostat with
> > the semi-isotropic coupling solved most of these problems. After the
> > structure is equilibrated , one can switch back to the MC barostat if
> > desired.
> >
> > I’m not sure if this will solve the problem but could be worth giving it
> a
> > try.
> >
> > Best,
> > Batuhan
> >
> > > On 3 Jun 2020, at 15:17, David A Case <david.case.rutgers.edu> wrote:
> > >
> > > On Wed, Jun 03, 2020, Dave Cerutti wrote:
> > >
> > >> If you could provide a link to the archive where that advice is
> found, I
> > >> might be able to comment more, but for now I looked through the manual
> > and
> > >> found the "baroscalingdir" keyword, which is there to control the
> > >> directions along which the box stretches during constant pressure
> > >> dynamics. This keyword is applicable with the Monte-Carlo barostat
> and
> > >> anisotropic pressure control. Since semi-isotropic is just a special
> > case
> > >> of anisotropic, I don't see a reason you cannot combine ntp=3 and
> > >> barostat=2. The detailed balance condition on the MC barostat is
> > >> important, so someone may have made a note of DtB considerations if
> the
> > box
> > >> does not stretch isotropically. However, if baroscalingdir can be an
> > >> option, any caveats are either solved or at least one person who
> > programmed
> > >> the MC barostat thinks it's OK.
> > >>
> > >>>
> > >>> I've also read, back in the Amber forums, that when you use ntp=3,
> you
> > >>> should use barostat=1. Is this true?
> > >>>
> > >
> > > Please check with Prof. Wonpil Im at Lehigh University. I think I
> > > remember that he has looked at this problem, but I'm not sure if this
> > > has been published. Other Amber users running membrane simulations
> with
> > > Amber force fields and the MC barostat might want to comment on their
> > > experience.
> > >
> > > ....dac
> > >
> > >
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org <mailto:AMBER.ambermd.org>
> > > http://lists.ambermd.org/mailman/listinfo/amber <
> > http://lists.ambermd.org/mailman/listinfo/amber>
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 04 2020 - 20:30:02 PDT
Custom Search