A tested energy function for protein structure prediction

PHYS 423

Matthew S. Lin, mattlin@berkeley.edu, UCSF & UCB Joint Graduate Group in Bioengineering, University of California Berkeley, Berkeley, CA 94720 and Teresa Head-Gordon, TLHead-Gordon@lbl.gov, Department of Bioengineering, University of California, Berkeley, Berkeley, CA 94720.
The protein structure prediction problem is to predict the three-dimensional topology of the native state of a protein given its sequence of amino acids. Besides a global optimization method, a scoring function that can effectively distinguish the native and the native-like structures from misfolded decoys is needed. Here we propose a new scoring function based on the AMBER force field in addition with the Generalized Born solvent model (AMBER/GB), and descriptions of hydrophobic hydration solvation forces (HS). Our research group has studied a different but critical influence of aqueous solvent on protein conformation, namely hydrophobic interactions, using both experimental solution scattering and computer simulation. We believe that our studies support the view that small length scale hydration physics is operative for the folding and stabilization of globular proteins. We test our energy function (AMBER/GB/HS) and AMBER/GB/SA on the protein sets from the EMBL, “Decoys ‘R' Us”, and Baker decoy database. AMBER/GB/HS outperforms AMBER/GB/SA on 95% of the 80 test proteins. It demonstrates that the HS term models the hydrophobic interactions in proteins better than the SA model.

Poster Session
7:30 PM-10:00 PM, Wednesday, 13 September 2006 Moscone Center -- Hall D, Poster

Division of Physical Chemistry

The 232nd ACS National Meeting, San Francisco, CA, September 10-14, 2006