Structure prediction for CASP7 targets using extensive all-atom

Rhiju Das,Bin Qian,Srivatsan Raman,Robert Vernon,James Thompson,Philip Bradley,Sagar Khare,Michael D. Tyka,Divya Bhat,Dylan Chivian,David E. Kim,William H. Sheffler,Lars Malmstrom,Andrew M. Wollacott,Chu Wang,Ingemar Andre,David Baker
2007-01-01
Abstract:We describe predictions made using the Rosetta structure prediction methodology for both template- based modeling and free modeling categories in the Seventh Critical Assessment of Techniques for Pro- tein Structure Prediction. For the first time, aggressive sampling and all-atom refinement could be car- ried out for the majority of targets, an advance enabled by the Roset- ta@home distributed computing network. Template-based modeling predictions using an iterative refine- ment algorithm improved over the best existing templates for the ma- jority of proteins with less than 200 residues. Free modeling methods gave near-atomic accuracy predic- tions for several targets under 100 residues from all secondary struc- ture classes. These results indicate that refinement with an all-atom energy function, although computa- tionally expensive, is a powerful method for obtaining accurate structure predictions.
What problem does this paper attempt to address?