Protein structure from scratch?

Arne Elofsson arne at hodgkin.mbi.ucla.edu
Thu Aug 11 13:49:07 EST 1994


In article <32djkp$hbs at charm.magnus.acs.ohio-state.edu>, dopearso at magnus.acs.ohio-state.edu (Douglas C Pearson) writes:
Path: news.mic.ucla.edu!unixg.ubc.ca!nntp.cs.ubc.ca!newsxfer.itd.umich.edu!news.cic.net!magnus.acs.ohio-state.edu!dopearso
From: dopearso at magnus.acs.ohio-state.edu (Douglas C Pearson)
Newsgroups: bionet.molec-model
Subject: Re: Protein structure from scratch?
Date: 11 Aug 1994 16:29:13 GMT
Organization: The Ohio State University
Lines: 23
Message-ID: <32djkp$hbs at charm.magnus.acs.ohio-state.edu>
References: <170097098S86.FJVAND00 at ukcc.uky.edu> <3287u5$7ki at charm.magnus.acs.o
NNTP-Posting-Host: beauty.magnus.acs.ohio-state.edu

>Floeckner Hannes <floeckn at wst.edvz.sbg.ac.at> wrote:
>>I wonder what you mean with '65% accurate model'. This "percentage correct"
>>measure is, as far as I know, always used in connection with algorithms for
>>the prediction of secondary structure (like the EMBL prediction server, xy%
>>of the residues assigned the correct secondary structure). But these secondary
>>structure predictions don't give a 3D model of the protein.
Fred Cohen showed a long time ago that packing of a four helical boundle 

1. Presnell SR; Cohen FE.
     Topological distribution of four-alpha-helix bundles.
   Proceedings of the National Academy of Sciences of the United States of
   America, 1989 Sep, 86(17):6592-6.

and several other references

>>By the way, I don't think 65% is a good result.
>
Rost and sander showed that proteins are approximately 82% correct
(within a family) So it is not that bad.

>i believe the connection with secondary structure is correct (though i'd be 
>happy if somebody could confirm that for me...i'm just a second-year grad 
>student, for pete's sake!).  i agree, that's not the best model in the world,
>but if it's a choice between knowing something about a protein's secondary/
>tertiary structure and knowing nothing, what are you going to take?
>this type of modeling is good as a starting point.  nothing more.>

You really need to find a protein with some homology to get some idea
of the structure. Sec.Str itself is hardly good enough.

arne

--
******************************************************************************************
**		Arne Elofsson (arne at hodgkin.mbi.ucla.edu)				**
**		Molecular Biology Institute, UCLA					**
**		405 Hilgard Avenue, Los Angeles						**
**		90024-1570 California, USA						**
**		tel: +1-(310)-825-1402  Fax: +1-(310)-206-3914				**
**		Home: +1-(310)-268-8006 SWE: +46-8-306129				**
******************************************************************************************



More information about the Molmodel mailing list