How to compare Protein folding success

John Kuszewski johnk at spasm.niddk.nih.gov
Mon May 8 14:53:29 EST 1995


In article <arneznecz8ususkm at hodgkin.mbi.ucla.edu>, arne at hodgkin.mbi.ucla.edu (Arne Elofsson) writes:
|> In article <950508105245.2020b39f at scri.fsu.edu> STRELETS at SCRI.FSU.EDU ("VICTOR B. STRELETS") writes:
|> 
|> 
|> > Almoust everybody in this area is talking about
|> > LOCAL folding first (and if GR slides sequence
|> > on 50aa segments first, he is doing local folding 
|> > too) and all fine and weird art next. What type
|> > of tests will be appropriate to test local folding
|> > algorithms? In this vase you don't have structure
|> > in whole, RMSD is inapplicable etc... Any idea?
|> 
|> 
|> An easy way would be to just count secondary structure 
|> prediction from these methods. If they really do better
|> than sec.str. predicition then it gets tougher to judge.
|> 
|> RMSD is bad, maybee contactmaps are better. Unfortunately
|> do I not know anything better than looking at the structures 
|> (or have a crystallographer to look at them).
|> 

"Local" predictions (in the sense of predicting tertiary strucutre
or simulating the folding of of 50-residue or so fragments) 
is not necessarily the physically incorrect thing to do.

By looking at folding/unfolding transitions using calorimetry,
Privalov has postulated that proteins fold as a series of 
relatively independent two-state reactions of sections 
50-100 residues long.

If you look at structures, he's probably right--very few sections
of really contiguous structure (ie., a large spacial concentration 
of specific contacts) that are larger than 100 residues exist 
in the database.

(Perhaps I'm not so sorry for mentioning the new approaches
to predicting structure.  It sure has stimulated a great
deal of discussion.  

The talk I went to, in case anyone cares, was not 
"priviledged information" in any way.  However, since
he has given talks at other places in the past few months 
and I was the only one to mention it here, my gut reaction 
is to hold off on the details.  It won't hurt anyone too 
much, since the paper should appear for the world to see
in a few months.)
-- 
                                   _____________
                                   |        ___/_
                                   |        |/  /
                                   --  /\  //  /--
                                   ||  ||  /  /||
                                   ||  || /  / ||
                                   ||  ||/  /  ||
John Kuszewski                     ||  |/  /|  ||      
johnk at spasm.niddk.nih.gov          ||  /  /||  ||
                                   \/ /  / ||  \/
that's MISTER protein G to you!     |/__/|      |
                                      /_________|

"Biophysics has driven me to an attitude of apocalyptic doom"
   --Frank Delaglio



More information about the Proteins mailing list