Modeling Structure By Homology

[Francois Jeanmougin]

In article <4reua6$ab3 at mawny.microagewny.com>,
	dpeaston at Microagewny.com (Douglas Easton) writes:
>I am currently modeling the structue of a protein using an X-ray structure for 
>a protein, distantly homologous in sequence and in function. I am predicting 
>the secondary structure using ssp, nnsp, GOR etc. And alligning  sequence 
>homologies and the real and predicted secondary structures. The results seem
>neat but my confidence in the approach is a bit shakey.
>
>Does anyone have experience in this?, Know of any particularly good papers?
>Can you carry this analysis to the tertiary structure level?
>
>Thanks in advance,
>-Doug Easton 

	Hi,
	Depends of what precision you need. To build protein using an 
homologous crystalized one, I first make a multiple alignment using ALL
the homologous proteins (all the family). For this I use ClustalW. It is
important to have a multiple alignment, because their  are more informations in.
	I check the quality of the alignment by using phd secondary structure
prediction (http://www.embl-heidelberg.de/predictprotein/phd_pred.html). 
	the prediction should match "about" the known structure.

	Then I use WhatIf, that have a whole module of homology building.

	After that, there is quite a work on ramachadran plot, and
sometimes a little bit of minimization (using the CHARMM module of Quanta).

	Then you have a probable structure, and can look for critical point
of mutations.

	Well, all of that needs to "eat" some manuals, and one or two
months of full time work (for debutant, less for crystallograph, but I'm not).

					hope this helps.

							Francois.

P.S.: Note that the critical step is the multiple alignment...
-- 
Francois Jeanmougin
Service de bioinformatique / bioinformatics service
IGBMC 
BP 163 
67404 Illkirch France
tel : (33) 88 65 32 71
e-mail : jeanmougin at igbmc.u-strasbg.fr

Les petits pois sont rouges (STTELLLA)