[MMTK] ENM normal mode analysis question

Konrad Hinsen research at khinsen.fastmail.net
Mon Jun 3 05:02:30 UTC 2013

Liz Kellogg writes:

 > thanks for the response! I am considering three different structures of 
 > the same protein sequence. The structures are very similar to one 
 > another, around 1 Angstrom rmsd, and the structures are large, around 
 > 850 aa.

OK, so you are looking for pretty small effects.

 > When I say "resolving" the structural changes, I mean this: Because the 
 > structures are so similar (yet have visible shifts when I align one on 
 > the others) it's difficult to see what is changing. I was hoping normal 
 > mode analysis could decompose the structural changes into something that 
 > is easier to parse visually (i.e. look at the structural differences 
 > between the three).

Normal mode analysis may help with this, but that's not guaranteed.
If the conformational changes you see turn out to be described by one
or two modes, then this will provide a much simplified description. If
the changes require ten modes for a good description, you will just have
converted one messy picture into a different one.

For a first visual analysis, I recommend using either my DomainFinder
tool (http://dirac.cnrs-orleans.fr/DomainFinder/) or the WEBnm@ server
in Bergen (http://apps.cbu.uib.no/webnma/home) on pairs of structures.
Both are based on MMTK's normal mode analysis, and use very similar
methods for comparing structures. Have a look at the deformation
energy profiles, which show you which parts of your protein remain
rigid in a structural change and which are deformed.

If you see large rigid areas separated by localized flexible ones,
then you have domain motions and normal mode analysis will be very
useful for providing a description. If the deformations are of roughly
the same amount everywhere, then it's unlikely that normal modes are
the right approach for you.

You can also look at the file Examples/Proteins/analysis_calpha.py
in the MMTK distribution to see how more advanced configuration
analysis methods can be realized.

 > Also, the three structures represent three distinct states. Is it 
 > possible to sample the conformational transition from one state to 
 > another with normal modes? (this question here is how does one structure 
 > turn into the other?)

You can create an interpolation between your structures using normal modes.
This has been proposed (and done) several times in the literature. The
big problem is that it's impossible to verify if such an interpolation
has anything to do with the real transition path.

Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: research AT khinsen DOT fastmail DOT net
ORCID: http://orcid.org/0000-0003-0330-9428
Twitter: @khinsen

More information about the mmtk mailing list