How to compare Protein folding success

Arne Elofsson arne at
Sun May 7 17:42:16 EST 1995

I do not really knows if this belongs here, but...

This is my own thoughts about this problem that I
think should be adressed before the next prediction 
meeting. After listening to the talks, reading some
of the manuscripts and so so on. I have realised that
even such a meeting is not guaranteed to be completely
fair if the format is not standardized. Otherwise it will end up
beeing very much a question about how much time you spend on it.

My suggestions:
1) Homology modelling: Submission of one PDB file.
2) Fold recognition: Submission of the list of Z-scores
   (or whatever energy you use) over folds, AND the
   fold you think it is. Nothing else. (There were some
   cases were some groups submitted not their top-ranking
   hits but the hit that made sense). Folds should probably
   be defined after the SCOP database. Finally should
   an alignment between your choice be presented.
3) Ab initio, Submission of one PDB file.
4) Secondary structure prediction. An assignment file.

The submissions should also be accomplished with a good
description of the methods. And preferably be made available 
to everyone a few weeks before the meeting.

This would put lot of the so called ab-initio work
into the fold recognition area but I think it belongs
there. I can not see any fundamental difference between
Tim Hubbard/Sean Eddy's work and David Jones'. The results
is more or less the same. You get a fold and an alignment.

my $0.02 


               From: Arne Elofsson
         Email: arne at   			

More information about the Proteins mailing list