Hi Sajeewa,
You've got most of it correct.
1) Yes, if all the r1,r2,r3,r4 values are different the potential has a
flat bottom, rather than being harmonic which is needed for umbrella
sampling (at least standard umbrella sampling where you are going to use
WHAM to build back the potential energy landscape). So yes, r2 and r3 need
to have the same value in order to make a harmonic potential. Technically,
a harmonic potential is only defined by its center and a force constant.
This means that r1 and r4 are not really part of that. You have two
options for dealing with them:
A) make them far enough from the center (r2,r3) that you are confident your
system will never venture that far, therefore your system will always see a
harmonic well. This is not really ideal, because your assumption might be
wrong, and then you can't use that simulation reliably for analysis.
B) set them to something simple (r1 = 0, r4 = r3 + 1), and then change some
variable (can't remember it's name right now, maybe "iat"?) so that the
shape of the potential remains harmonic beyond r1 and r4, rather than
becoming linear or flat (the other options for that variable). This is the
best option, as it now means that by defining r2 = r3 = harmonic_center,
you get the correct potential that umbrella sampling needs, regardless of
what happens.
Keep in mind for all this r1,r2,r3,r4 business, that the potential is setup
as something more complicated than a harmonic potential, because AMBER was
initially used to refine protein structural models (Assisted Model Building
and Energy Refinement or something like that). So if you have NMR data
that tells you Atom1 and Atom2 should be between 5 and 7 angstroms, you
want any value between 5 and 7 those to have no penalty (r2 =5, r3 =7), and
apply a penalty if they go beyond that (e.g. r1 = r2 - 2 = 3, r4 = r3 + 2 =
9), but if the system seems to want to go even further away from your 5 to
7 range, don't continue giving it a harsh harmonic penalty, but rather a
nicer linear or flat one, because maybe your NMR data is not reliable for
that pair of atoms.
2) None of the atoms involved in your reaction coordinate should be fixed
if you want to properly do umbrella sampling, otherwise you are applying an
additional bias that you can't undo later during analysis with WHAM or
equivalent method. So, the distance restraint for instance is as you say,
between Atom1 and Atom2 or whatever, but both of them will respond to that
force.
An additional note here, if you can (not sure what is available in AMBER12
or whatever you are using), it's better to define groups of atoms and use
their center mass, so for a protein ligand system, the distance might be
between the center of mass of several heavy atoms in the protein binding
pocket, and the center of mass of all the ligand heavy atoms. With enough
sampling this is irrelevant, but usually one doesn't have quite enough, and
this can help reduce noise due to other degrees of freedom like rotations
that aren't being explicitly biased against.
In response to the last thing you wrote. The distance restraint, that is,
the harmonic potential, which will be centered at (r2 +r3)/2 and is the
distance between Atom1 and Atom2, not anything to do with absolute
coordinates (I'm not sure what you meant by what you wrote). So, although
r1 and r4 are irrelevant for umbrella sampling, r1 is a value that would be
a shorter distance, and r4 would be a longer distance between your two
atoms).
Hope that helps,
~Aron
On Sun, Nov 4, 2012 at 11:32 PM, Sajeewa Pemasinghe <sajeewasp.gmail.com>wrote:
> Hi all,
>
> I am new to umbrella sampling and this is the first time I am doing this. I
> have some questions
>
> 1) The shape of the biasing potential is more or less "U" shaped if we give
> four different values for r1,r2,r3 and r4 but usually to approximate a
> harmonic shape we give the same value to r2 and r3. Am I correct?
>
>
> 2) When we give distance restraints we usually write (atom#1 is fixed and I
> am looking at the change in distance of atom#2 from atom#1)
>
> &rst iat=1,2, r1=5, r2=10, r3=10, r4=15, rk2 = 32.83, rk3 = 32.83, /
>
> In amber tutorial http://ambermd.org/tutorials/advanced/tutorial4/ it says
> that
>
> "Conventionally, the violation energy is a well with a square bottom
> between *r2* and *r3*, with parabolic sides out to a defined distance (*r1*
> and *r4* for lower and upper bounds, respectively), and linear sides
> beyond that distance"
>
> Does the above mean that r1 is the distance to the left (towards zero) from
> r2 and r4 is the distance to the right from r3 OR are all distances
> measured from my reference atom (atom#1 in my case)?
>
> I really appreciate your help.
>
> Thank you
>
> Sajeewa Dewage
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Aron Broom M.Sc
PhD Student
Department of Chemistry
University of Waterloo
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Nov 04 2012 - 22:30:03 PST
