Re: metadynamics combined with PCV

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Fri Jun 03 2011 - 12:28:53 CDT

Hello Anurag, the path collective variables (PCV) may be implemented
in the near future, but at present it isn't. I'm currently working on
other things.

My guess is that unless you approximate the exponential functions with
polynomials (which are implemented) you won't be able to define PCVs.
And you can't use Tcl syntax to define a function of the existing
collective variables. If you're extremely interested in using it and
are fluent in C++ , we can communicate in private on how to start with
either approach (native PCVs or Tcl-based).

You may also want to have a look at PLUMED: it is not part of native
NAMD and you wouldn't have certain features that you otherwise get
with colvars, but it is more in sync with the variety of more recent
metadynamics-based methods that are being developed by the Parrinello
group.

Giacomo

On Fri, Jun 3, 2011 at 7:14 PM, Anurag Sethi <anurag.sethi_at_gmail.com> wrote:
> Hi,
>
> I am trying to run metadynamics with NAMD 2.8b1 to study conformational
> changes associated with allostery in HIV envelope proteins.  I would like to
> use the approach suggested by the Parrinello group in which they combine
> path-based methods with the approach of collective variables and apply
> metadynamics to study coformational changes in small peptides (Branduardi,
> et al., J. Chem. Phys., 126:054103 (2007)) and kinases (Berteotti, et al.,
> J. Am. Chem. Soc., 131:244-150 (2009)).  In this method, they identify a
> putative path by generating a morphed path from the initial conformation (A)
> to the final conformation (B).  After this, they define two path collective
> variables (PCVs).  One of these PCVs (s in their notation) measures the
> "progress" along the putative path while the other PCV (z in their notation)
> measures the "distance" of the conformation from the putative path.  Using
> this method, they are able to find multiple pathways for conformational
> changes using metadynamics (in combination with the nudged elastic band in
> their case).
>
> Both these PCVs use exponential functions of RMSD of the protein from
> multiple conformations (defined by the putative path) of the protein.  It is
> equations 1 and 2 in the 2007 paper referenced above.  I have read the
> documentation for the colvars module in NAMD and see that RMSD is one of the
> colvars defined in NAMD.  However, I also see that polynomial functions of
> these colvars can be used in NAMD directly.  Is there any way for me to use
> tcl along with NAMD to define these PCVs and run metadynamics on these PCVs?
>
> Thanks for your help,
> Anurag
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:57:14 CST