On Thu, Dec 25, 2014 at 10:46 PM, Sun <sunbintyy.163.com> wrote:
> Dear all,
>
> I studied the disassociation process of a ligand depart from an enzyme's
> active site by using the PMF method. The umbrella sampling and Weighted
> Histogram Analysis Method (WHAM) were used. However, when I used the WHAM
> to get final PMFs of this process, I found that the tolerance value, a
> parameter which specifies the convergence criteria of the WHAM calculation,
> influenced the free energy curve abnormally ( see the following figure).
>
You didn't actually attach any figure.
> In general sense, the smaller the tolerance, the more accurate the free
> energy will be. In my case, when tolerance was set to 0.1, the calculated
> binding free energy agreed with the experimental binding free energy quit
> well. However, when the tolerance became smaller, the deviation between
> calculated and experimental binding energy became larger, also the free
> energy curve became abnormal.
>
> So could anyone give me some advice about the possible factors that may
> cause this phenomenon ?
>
Are you sure that you have sufficient overlap between the various
windows? That is, do the geometric distributions (e.g., distance, angle,
or torsions) along the PMF dimension for each window overlap with their
adjacent windows? If not, WHAM will not be able to "stitch" the various
windows back together and you will get nonsensical results. You may need
to add more windows to ensure that you get good overlap. Not all methods
require windows to overlap (like the variational free energy profile
method), but WHAM does.
See the end of the umbrella sampling tutorial, section 2, for a brief
discussion about this:
http://ambermd.org/tutorials/advanced/tutorial17/section2.htm
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Dec 30 2014 - 15:30:03 PST
