Re: PCA Analysis of CGMD

From: Joshua Adelman (jadelman_at_berkeley.edu)
Date: Fri Apr 10 2009 - 10:20:04 CDT

Hi Anirban,

As long as you have a pdb for your system, you should be able to do
this. The analysis tools will probably complain that you don't have
connected chains, but that shouldn't effect the calculation. You might
have to edit the aminoacids.dat file if gromacs complains, and
potentially a file with the bead masses. I'm not sure since I haven't
done this with a CG system. As far as Rogan's comment about making an
index file -- you shouldn't have to mess around with make_ndx since
you'll get a prompt that asks you what subset of atoms you want to
use. I assume you'll just run PCA on the entire set of CG atoms. If
not you can create an ndx file by hand, by having something like:
[ Subsystem Name ] <------------ arbitrary identifier
1 3 5 7 9 11 <----------- serial numbers of atoms you
want in the system to analyze

you can define several groups if you want to do PCA on a different
group of atoms than the ones you do the initial superposition with.

Josh

On Apr 10, 2009, at 2:08 AM, Anirban Ghosh wrote:

> Hello Josh,
>
> Thanks a lot for the reply. I think the Gromacs idea is a good one.
> But since I have a CG model with beads (and not amino acids), will
> it be possible to use g_covar (which calculates eigen vectors and
> eigen-values on the basis of C-alpha) to do this?
>
> Thanks,
>
>
> Anirban Ghosh
> Grade Based Engineer
> Bioinformatics Team
> Scientific & Engineering Computing Group
> Centre for Development of Advanced Computing
> Pune, India
>
> From: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] On
> Behalf Of Joshua Adelman
> Sent: Thursday, April 09, 2009 8:58 PM
> To: Anirban Ghosh
> Cc: namd-l_at_ks.uiuc.edu
> Subject: Re: namd-l: PCA Analysis of CGMD
>
> Hi Anirban,
>
> There are a lot of options out there for doing this from writing
> your own code (for example using the MatDCD package to load the
> coordinates into matlab), or using a number of other available
> packages. The way I like to do it is to use catdcd to convert my
> trajectories into .trr format and then use the tools available in
> Gromacs (g_covar to calculate the eigenval/vecs and g_anaeig to
> project the trajectories onto particular modes and to generate pdb
> files to visualize them), to do PCA.
>
> Josh
>
>
>
> On Apr 9, 2009, at 5:23 AM, Anirban Ghosh wrote:
>
>
> Hi ALL,
>
> I have done a CGMD simulation of a three chain protein system using
> NAMD. Is it possible to do a Principle Component Analysis (PCA) of
> this CGMD run? If yes, then how? Any suggestion is welcome.
>
> Regards,
>
>
> Anirban Ghosh
> Grade Based Engineer
> Bioinformatics Team
> Scientific & Engineering Computing Group
> Centre for Development of Advanced Computing
> Pune, India
>
>

------------------------------------------------------------------------------------------------------
Joshua L. Adelman
Biophysics Graduate Group Lab: 510.643.2159
218 Wellman Hall Fax: 510.642.7428
University of California, Berkeley http://nature.berkeley.edu/~jadelman
Berkeley, CA 94720 USA jadelman_at_berkeley.edu
------------------------------------------------------------------------------------------------------

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:35 CST