Re: [AMBER] On the choosing of coordinate for the umbrella sampling

From: Brian Radak <>
Date: Thu, 5 Jun 2014 09:37:29 -0400

As far as I know, AMBER is only capable of utilizing simple geometric
coordinates (distances, angles, dihedrals, *etc.*) for harmonically biased
umbrella sampling. Slightly more complicated coordinates are available, but
frequently *only* *in sander*. If I'm not mistaken, there are significant
performance issues with arbitrary restraints and domain decomposition (the
main source of performance boost in pmemd). More generally applicable (and
high performance) coordinates are also available via accelerated MD (which
I would argue is just a particular form of umbrella sampling) or by
directly augmenting the prmtop file (*e.g. *using

Fortunately, modern PMF reconstruction methods generally do not require
that the bias coordinate(s) correspond to the reaction/progress
coordinate(s) in any way, shape, or form. The only requirement is that the
biases provide adequate sampling of the desired phase space (with all of
the statistical ambiguities that implies). Clearly it is often the case
that it is convenient for these coordinates to be one and the same, but
that is by no means a restriction.


On Thu, Jun 5, 2014 at 7:45 AM, Jason Swails <> wrote:

> On Thu, Jun 5, 2014 at 7:03 AM, James Starlight <>
> wrote:
> > Dear Amber users!
> >
> >
> >
> > I'd like to compute of PMF along some coordinate specified transition of
> my
> > protein from the inactive R to the active R* form. I'll be thankful for
> > someone who could provide me with some tutorial with the guide for
> > calculations of the lowest minimum transition pathway between R and R*
> > provided in two pdbs and using it for the further PMF as some collective
> > variable (e.g combination of eigenvectors or something like this).
> >
> ​This sounds like another question that is best answered by literature
> searches and experimentation.
> The art of PMF calculations lies in picking the reaction coordinate (RC)
> about which the PMF is computed, and there is no tried-and-true recipe
> (that I'm aware of) for picking the ​RC for umbrella sampling. You
> typically need some kind of prior knowledge about how the system behaves in
> order to pick a PMF. And even then, PMFs are far from unique and different
> choices accomplish different goals (compare a 2-D PMF of 2 distances with a
> 1-D PMF made up of a linear combination of those same 2 distances).
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> AMBER mailing list

================================ Current Address =======================
 Brian Radak                                             :     BioMaPS
Institute for Quantitative Biology
 PhD candidate - York Research Group       :     Rutgers, The State
University of New Jersey
 University of Minnesota - Twin Cities         :     Center for Integrative
Proteomics Room 308
 Graduate Program in Chemical Physics     :     174 Frelinghuysen Road,
 Department of Chemistry                          :     Piscataway, NJ
08854-8066                                 :
Sorry for the multiple e-mail addresses, just use the institute appropriate
AMBER mailing list
Received on Thu Jun 05 2014 - 07:00:03 PDT
Custom Search