Building models in PDB format!

Andrej Sali sali at rockvax.rockefeller.edu
Tue Nov 26 13:17:56 EST 1996


Raghavender Pillutla wrote:
> 
> Hello Everybody,
> 
> I am trying to build shape models of some plasma proteins. Basically,
> I am just interestedin the overall shape of the protein. I would like
> to know what all information I need to know get the model in a PDB
> format (the .pdb extension). Also, I would like to know whether I need
> a specific software for building the model. Can someone help me?
> Thanks in Advance,
> 
> Raghav

If you can find a homologus known structure that does not have too many
too long gaps relative to the sequence you are interested in, you can
just take the known homologous strucure as an approximation of the shape
of the related sequence. In other words, a homology model is not likely
to give you a more accurate result for two reasons: (1) because the
surface sidechains are modeled inaccurately and (2) because the surface
sidechains frequently do not have a fixed conformation anyway. This is
so because sidechain conformations are surprisingly important in
determining the shape of a protein domain. They are so important because
the size of the longer sidechains (cca 5A) is not that much smaller than
the radius of most of the domains (cca 125 residues). A simple
experiment is this: take two homologous proteins with known structure
and about 50% sequence identity from PDB and display their surfaces
(e.g., by GRASP): you will be surprised how different their shapes are.

If you still want to do homology modeling, there several sites/programs:
SWISS-MODEL, MODELLER at http://guitar.rockefeller.edu, WHATIF,
COMPOSER, and others. 

Best, Andrej

-- 

Andrej Sali, Assistant Professor
The Rockefeller University, 1230 York Avenue, New York, NY 10021-6399
tel +1 212 327 7550; lab +1 212 327 7206 ; fax +1 212 327 7540
e-mail sali at rockvax.rockefeller.edu; http://guitar.rockefeller.edu



More information about the Proteins mailing list