On Sat, May 8, 2010 at 11:45 AM, Jeffrey <jeffry20072008.yahoo.cn> wrote:
> Many thanks for the detailed explanation.
>
> I think the theoretical H(lambda) and the one used by AMBER is different.
> The bond, angle and dihedral terms in the theoretical H(lambda) should not
> be independent to lambda.
>
True, because the separation of the hamiltonian into bond, angle, dihedral,
eel, and vdw is not an orthogonal separation, so to the extent that the
bonds/angles/dihedrals depend on charge/vdw changes, those terms will
contribute to DV/DL. Their effect will probably be fairly minimal (and the
differences between the model and real systems will probably also cancel to
a certain extent).
I believe that the original comment was aimed at the fact that the
contribution of the few terms directly related to the disappearing proton
will not depend on Lambda (to a great extent), and should cancel between the
model and real residues, anyway. Thus, disappearing that extra bond term
and the very few extra angle and dihedral terms along with the charges and
van der waals interactions (which shouldn't be an issue for HO atoms anyway,
since they have no VDW radius to begin with) is unnecessary. DV/DL will be
dominated by the other terms, anyway. The other bond, angle, dihedrals are
not *ignored*, per se, but the values of their force constants and such do
not depend on Lambda (since they are the same in both systems). I believe
DV/DL is just calculated as a difference between the energies computed with
both hamiltonians (so the closer together the lambdas are, the more the
slope approaches a true derivative). Ultimately, that handful of extra
terms arising due to the use of the dummy atom where there should be none is
insignificant and cancels to a great extent, anyway, so it's ignored.
I don't think my previous post quite got across what I was trying to, and
may have been slightly off the mark. I think I did a better job here
(though perhaps still confusing and misworded on occasion).
All the best,
Jason
> (1) Suppose that we used the model system in tutorial for example, the
> Hamiltonian under which TI samples the conformations should be
>
> H(Lambda) = Lambda * H1 + (1 - Lambda) * H2
> with H1 is the Hamiltonian of R-COOH and H2 of R-COO-
>
> In addition to the energy terms included in H2, the H1 contains extra bond,
> angle, dihedral, ele, vdw terms involving in the H atom of hydroxyl group.
> Thus, after breaking the H(lambda) into component parts, the angle, dihedral
> and torsional terms involving in the H atom will still contain Lambda and
> their derivative should also contribute to the DV/DL.
>
> If we calculate the Delta Delta G btw the model and protein system, this
> contribution arising from local conformational can be cancelled anyway.
> However, the contribution of these terms are directly included into the
> DeltaG of either model or protein system.
>
> --
> (2) But if we construct a "dummy" atom in H2 and don't remove the angle,
> bond, and dihedral terms involving in the dummy atom in H2, which is adopted
> by the tutorial, all the bond, angle and dihedral terms in H(lambda) will be
> independent to lambda. In this case, these terms will not contribute to the
> free energy calculation in either model system or protein system. In this
> case, the contribution of these terms are not directly included into the
> DeltaG of either model or protein system, but would effect the DeltaG value
> during the sampling process. When DeltaDeltaG is calculated btw model and
> protein system, these contributions can be cancelled under the hypothesis
> that the average sampling behaviors contributed by these terms in model and
> protein are identical.
>
> Thus, the cancellation of these terms in DeltaDeltaG is not the same as in
> (1).
> Maybe the AMBER adopts the second scheme for convenience. But I don't think
> these terms in theorectical H(lambda) (as in (1)) should be independent
> to Lambda.
>
> I am not sure whether I understand this question correctly or not in this
> way and thus request further elucidation.
>
> Many thanks.
>
> Jeffrey Yang
>
>
> --- 10年5月8日，周六, Jason Swails <jason.swails.gmail.com> 写道：
>
>
> 发件人: Jason Swails <jason.swails.gmail.com>
> 主题: Re: [AMBER] How to disable the VDW interaction?
> 收件人: "AMBER Mailing List" <amber.ambermd.org>
> 日期: 2010年5月8日,周六,下午9:17
>
>
> Hello,
>
> On Sat, May 8, 2010 at 8:44 AM, Jeffrey <jeffry20072008.yahoo.cn> wrote:
>
> > Thanks for your clarification. I have two other relating questions:
> > 1. If I make a new residue file with a different atom
> > type, which parameters for bond, angle, and dihedral involving the new
> atom
> > type should be used? Can these parameters be copied to the new atom type
> by
> > tleap when construting the prmtop file?
> >
>
> You will need to create an frcmod file for this. You can put the same
> parameters in the frcmod as were loaded by tleap for the old atom type.
> parmchk should show you all the parameters that you need to incorporate,
> and
> you can compare with $AMBERHOME/dat/leap/parm/parm99.dat (or whatever force
> field you're using) to get the proper parameters.
>
>
> > 2. In fact, I am not very clear about the statement "We don't need to
> worry
> > about removing the bonds, angles and dihedral terms as these are all
> > internal to the residue and so the average properties of these terms
> should
> > be identical in the model and protein systems. There is also no
> > contribution from these terms to the derivative of the potential with
> > respect to lambda." in the tutorial for pka calculation by TI method as
> in
> > http://ambermd.org/tutorials/advanced/tutorial6/section1.htm
> >
>
> When you're doing TI, the free energy change between two hamiltonians is
> calculated via the differential (dH/dL) where L is lambda. If you break
> the
> hamiltonian into its component parts, you get
>
> H = bond + angle + dihedral + eel + vdw + 1-4 interactions. Because the
> differential operator is linear, you can separately take the derivative of
> each term with respect to lambda. The bond, angle, and dihedral parameters
> are not changing as a function of Lambda, so those derivatives should be
> zero, since, with respect to Lambda, they're constant (and you're
> differentiating with respect to Lambda).
>
> The tutorial is saying this is a valid approach (i.e. it is not needed as a
> further refinement to change bond/angle/dihedral parameters) because they
> are only dependent on the local conformations adopted by the residue in
> question, and those local conformations should be distributed approximately
> equivalently between the model residue and the residue in the protein
> matrix. Thus, when you take DELTA DELTA G, they cancel out anyway (since
> TI
> requires sampling, which averages over all properties).
>
> All the best,
> Jason
>
> >
> > Suppose the following Hamiltonian is used,
> > H(lambda) = lambda * H1 + (1-lambda) * H2, which H1 and H2 correspond to
> > the protonated and unprotonated states, respectively.
> >
> > If the bond, angle and dihedral/torsion interactions involving
> > the disappeared hydrogen atom is not removed from the unprotonated state
> > (prmtop), the derivative of the H(lambda) with respect to lambda (H1-H2)
> > will also include the contribution from these terms. Why the tutorial
> says
> > that "There is also no contribution from these terms to the derivative of
> > the potential with respect to lambda." Do I understand it correctly?
> >
> > Any suggestion is appreciated.
> > Many thanks.
> >
> > Mingjun
> > --- 10年5月8日，周六, Carlos Simmerling <carlos.simmerling.gmail.com> 写道：
> >
> >
> > 发件人: Carlos Simmerling <carlos.simmerling.gmail.com>
> > 主题: Re: [AMBER] How to disable the VDW interaction?
> > 收件人: "AMBER Mailing List" <amber.ambermd.org>
> > 日期: 2010年5月8日,周六,下午6:01
> >
> >
> > 1) the RADII section is for the intrinsic radius for the implicit
> > solvation calculations. it is not vdw radius.
> > 2) the prmtop does not contain per-atom vdw parameters, only pointers
> > into the atom type arrays that hold the parameters. so, you can't
> > change 1 atom's parameters there, because you would have to change
> > them for all hydrogens of the same atom type.
> > 3) it seems to me that you need to go back to the actual residue
> > library file and determine the atom type- some of the hydrogen atoms
> > in Amber's force fields already have zero for the vdw well depth (you
> > want that, not zero for radius). you need to check and see what it is,
> > and if it isn't zero then you will need to make a new residue file
> > with a different atom type that has zero for vdw interaction. it's nI ot
> > as simple as modifying the prmtop.
> >
> > On Sat, May 8, 2010 at 2:51 AM, Jeffrey <jeffry20072008.yahoo.cn> wrote:
> > > Dear AMBER users,
> > >
> > > We are going to calculate the pka of a protonation of residue in a
> > protein model, following the tutorial at
> > http://ambermd.org/tutorials/advanced/tutorial6/. I disabled the vdw
> > interaction by setting the vdw radii of the hydrogen atom to 0.0 in the
> > prmtop file (the term under %FLAG RADII). But it is strange that the vdw
> > interaction didn't change w/o modification of the radii.
> > >
> > > I am not sure whether I did it correctly or not and so request your
> > suggestions.
> > >
> > > Many thanks.
> > >
> > > Jeffrey Yang
> > >
> > >
> > >
> > > _______________________________________________
> > > 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
> >
>
>
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Graduate Student
> 352-392-4032
> _______________________________________________
> 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
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat May 08 2010 - 16:30:02 PDT
