Re: Implicit solvent lipid bilayer simulation in NAMD.

From: Kenno Vanommeslaeghe (kvanomme_at_rx.umaryland.edu)
Date: Thu Sep 18 2014 - 15:39:49 CDT

Just so that we're all on the same page, there surely do exist PB- and
GB-based implicit solvent models that have the ability to mimic a membrane
(essentially by varying the implicit solvent properties in function of the
Z coordinate), and a few of them are in fact implemented in the CHARMM
program. However, these methods assume that the membrane itself is *also*
represented implicitly. I'm not sure I've ever seen an implicit solvent
simulation containing an explicit membrane, and I wouldn't expect this to
work well; there are several possible issues, among which the ones Jason
brought up.

These models are understood to be approximations, of course; when accuracy
is the first and foremost priority, one arguably should stay away from
implicit solvents altogether. Keeping that in mind, one can use
CHARMM-GUI's Implicit Solvent Modeller to generate input files:
http://www.charmm-gui.org/?doc=input/implicit
(the .doc files linked on that page have citations)

Focusing on the question at hand, I don't know whether any of these models
are implemented in NAMD. Your best guess is to, you know, read the fine
manual. Since an implicit membrane requires several parameters to be
defined (z-coordinates,...), it must be in the manual if it's supported at
all.

On 09/17/2014 03:27 PM, Subbarao Kanchi wrote:
> Hi Jason and Aron,
> I have read few recent papers where they
> have been used the distance dependent dielectric constant model instead of
> explicit solvent with CHARMM ff. But they have used the CHARMM md package.
> I really want to know whether the distance dependent dielectric
> implemented in NAMD and same time I am trying with the generalize born
> implicit solvent.
>
> Regards,
> Subbu.
>
> On Thu, Sep 18, 2014 at 12:24 AM, Aron Broom <broomsday_at_gmail.com
> <mailto:broomsday_at_gmail.com>> wrote:
>
> I think if you wanted to simulate how a protein for instance might
> behave if entirely constrained within a bilayer, but without actually
> using the lipids, there may be some work being done to develop
> implicit solvent methods where the solvent is a lipid-like thing
> (dissolving your protein in hexane for instance).
>
> There may also have been some success with using coarse-grained
> methods for doing things with lipid layers (particularly I think the
> NAMD supported MARTINI forcefield has been used for this), if you are
> trying to save on performance from the necessarily massive systems one
> gets when solvating a big bilayer patch.
>
> On Wed, Sep 17, 2014 at 2:47 PM, Jason Swails <jason.swails_at_gmail.com
> <mailto:jason.swails_at_gmail.com>> wrote:
>
>
>
> On Wed, Sep 17, 2014 at 2:41 PM, Subbarao Kanchi
> <ksubbu85_at_gmail.com <mailto:ksubbu85_at_gmail.com>> wrote:
>
> Hi Siva,
> Thank you for the reply. I read the namd manual
> and you are correct it is not compatible with PME. I have not
> used the PME in my input file.
>
>
> I don't see how GB can possibly work for lipid bilayer
> simulations. The only lipid simulations I've seen in recent work
> has used a periodic unit cell (and normally long-range
> electrostatics, like PME or RF or something).
>
> For one thing, GB represents an infinitely dilute system (since
> the solvent goes out to infinity in all directions and you have a
> finitely countable number of lipid molecules). Under such
> experimental conditions, you would expect the bilayer to
> _eventually_ dissolve and dissociate, so this approach seems
> doomed from the start (at least to me).
>
> For another thing, I'm not aware of any efforts to parametrize GB
> models for lipids -- probably because it wouldn't work well
> regardless of model parameters.
>
> Good luck,
> Jason
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
>
>
>
> --
> Aron Broom M.Sc
> PhD Student
> Department of Chemistry
> University of Waterloo
>
>

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:51 CST