[gmx-users] Re: Pull Code

[David L. Bostick]

Hi Justin,

I think I remember you posting something to the list about constraint
forces option not working for calculation of PMFs, but I'm not sure.  So,
what is the status of this part of the umbrella sampling code.  Does
constraing forces work?  If so, does it work in parallel also, or does only
the code currently in CVS have that capability?

Thanks,
David

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
David Bostick					Office: 262 Venable Hall
Dept. of Physics and Astronomy			Phone:  (919)962-0165
Program in Molecular and Cellular Biophysics
UNC-Chapel Hill
CB #3255 Phillips Hall				dbostick at physics.unc.edu
Chapel Hill, NC 27599	           		http://www.unc.edu/~dbostick
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

On Tue, 7 May 2002, Justin MacCallum wrote:

>
> Hi David,
>
> Do you have any idea what you expect the profile to look like?  Just so
> you know, g_wham outputs the probability distribution, not the free energy
> profile. You can convert the output to a free energy using xmgrace by
> evaluating y= -RT ln(y). I got what appears to be a reasonable profile
> using:
> 	g_wham -min -.54 -max .33 *.pdo -tol 0.001 -bins 250
>
> It looks fairly reasonable, but the free energies involved are rather
> large.
>
> You can also look at the histograms to see if your sampling is continuous
> using:
> 	xmgrace -nxy histo.xvg
>
>
> The version of g_wham I have here may be different than the one you have.
> I sort of have two versions floating around and have been working at
> merging them into CVS. Maybe I should do that sooner rather than later :),
> as there seem to be a number of people on the mailing list asking
> questions about the pull code.
>
> Justin
>
> ---------------------------------
> Justin MacCallum
> jlmaccal at ucalgary.ca
>
> Department of Biological Sciences
> University of Calgary
> Canada
> ---------------------------------
>
> On Mon, 6 May 2002, David L. Bostick wrote:
>
> > HI Justin,
> >
> > I have attached a tarball of 10 pdo files.  I manually gzipped them to run
> > g_wham, so they have the extension *.pdo.gz.pdo just so I could remember
> > that I had gzipped them.  Please let me know what happens..
> >
> > Thanks,
> > David
> >
> > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
> > David Bostick					Office: 262 Venable Hall
> > Dept. of Physics and Astronomy			Phone:  (919)962-0165
> > Program in Molecular and Cellular Biophysics
> > UNC-Chapel Hill
> > CB #3255 Phillips Hall				dbostick at physics.unc.edu
> > Chapel Hill, NC 27599	           		http://www.unc.edu/~dbostick
> > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> >
> > On Mon, 6 May 2002, Justin MacCallum wrote:
> >
> > >
> > > > Hi Justin,
> > > >
> > > > I have been doing umbrella sampling with a chloride ion moving through a
> > > > membrane protein in the z-dimension.  I am restraining it to the protein's
> > > > backbone in the z-dimension only.  So far I have sampled about 0.9 nm .  I
> > > > have used g_wham to do the free energy profile.
> > > >
> > > > It seems that g_wham is giving some very strange results:
> > > >
> > > > 1) I get different profiles depending on the order I type them on the input
> > > > line.
> > >
> > > That obviously seems very strange.
> > >
> > > >
> > > > 2) I get different profiles if I split the reaction coordinate into two
> > > > halves.  I'm not talking about magnitudes of free energy b/c it is
> > > > understood that this is all relative, but the profiles actually have peaks
> > > > in different places.
> > >
> > > This could just be due to a lack of sampling, but combined with #1,
> > > something seems to be wrong.
> > > >
> > > > 3) There is always only one maximum in the profile, and it is usually near
> > > > the beginning of the reaction coordinate.
> > > >
> > > > I'm not asking for support on this program b/c I understand that it is
> > > > developmental, but have you had this sort of problem?  You said you have
> > > > used g_wham for your umbrella sampling analyses.  What have your
> > > > experiences been?
> > > >
> > >
> > > Exactly which version of g_wham and the pull code are you using? I have
> > > been slowly working on the cvs version, but I'm actually using an older
> > > version for my runs. Could you possibly send me your pdo files, and I'll
> > > try to figure out what's wrong.  By the way, the pull code should improve
> > > dramatically in the next couple of months. I'm just in the middle of
> > > writing something up, and as soon as I'm done I plan to finish the pull
> > > code and add documentation, etc.
> > >
> > > Justin
> > >
> > >
> > >
> >
>
>