[gmx-developers] decoupling in free energy calculations for binding

[David Mobley]

Hi,

The specific thing we're trying to achieve here (in this case) is add a flat
bottom distance restraint between an atom in the ligand and a virtual site
in the center of a protein binding site. Not sure if that can be achieved
with the pull code?

We are also (separately) interested in having distance, angle, and dihedral
restraints between reference atoms in the ligand and reference atoms in the
protein. These restraints need to be lambda dependent, and ideally we would
like to be able to use them with decoupling (again, currently not possible).
Is the pull code flexible enough to be able to do angle and dihedral
restraints somehow? (There is a fairly specific set of restraints we need
for technical reasons -- one distance, two angles, three torsions. Other
forms won't get us what we need...).

Thanks.


On Thu, Jun 23, 2011 at 4:23 AM, Berk Hess <hess at cbr.su.se> wrote:

> **
> Hi,
>
> You can manually set up the decoupling, but that's very tedious and error
> prone.
> I think the best procedure would be to restrain using the pull code,
> but I don't know if that's currently flexible enough to cover most cases.
> If not, we should think about extend the functionality of the pull code.
>
> I though the decoupled state is described somewhere in the manual.
> It is a non-periodic state with pure Coulomb and LJ interactions without
> cut-offs
> (unless you use couple-intramol).
>
> Berk
>
>
> On 06/23/2011 04:46 AM, David Mobley wrote:
>
> Hi,
>
>  We're trying to do absolute binding free energy calculations using
> Michael Shirts' latest free energy code (to be included in 4.6 if I
> understand correctly; right now it's  a branch of 4.5). These require using
> a restraint between the protein and the ligand, which currently we're doing
> using virtual sites. I am interested in also doing these calculations using
> "decoupling", wherein internal interactions of the perturbed molecule (in
> this case the ligand) are retained. This involves something like the
> following in the mdp file:
>
>  couple-moltype            = MOL
> couple-lambda0            = vdw-coul
> couple-lambda1            = none
> couple-intramol           = no
>
>  if, for example, the ligand is a molecule named 'MOL'.
>
>  My question is this: Is there any way to get this to work when the ligand
> and the protein are part of the same "molecule"? Specifically, to have
> restraints betweeen the protein and the ligand (such as using a virtual
> site) which are NECESSARY for absolute binding free energy calculations, I
> must have the protein and ligand as part of the same molecule (unless
> there's a workaround I'm not aware of). But to get decoupling to work using
> the above I seem to need to have the protein and ligand as separate
> molecules, suggesting they are incompatible. Is there a workaround I'm not
> aware of?
>
>  Also, on a related note -- when decoupling is done, what is the end state
> for the decoupled object? Is it the decoupled object in gas phase in a
> periodic system (interacting with copies of itself), or in a nonperiodic
> system?
>
>  Thanks!
>
>  --
> David Mobley
> dmobley at gmail.com
> 504-383-3662
>
>
>
>
> --
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-developers-request at gromacs.org.
>



-- 
David Mobley
Assistant Professor
Department of Chemistry
University of New Orleans
dlmobley at uno.edu
ph. 504-383-3662
fax 504-280-6860






-- 
David Mobley
dmobley at gmail.com
504-383-3662
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20110624/a8e2707f/attachment.html>