Regensburg 2010 – scientific programme

Parts | Days | Selection | Search | Downloads | Help

BP: Fachverband Biologische Physik

BP 5: Posters: Biopolymers and Biomaterials

BP 5.28: Poster

Monday, March 22, 2010, 17:15–20:00, Poster B1

Selecting structure prediction candidates using sequence-derived structure profiles — •Katrin Wolff¹, Michele Vendruscolo², and Markus Porto¹ — ¹Institut für Festkörperphysik, TU Darmstadt, Germany — ²Department of Chemistry, University of Cambridge, UK

Selection of promising structure candidates for high-resolution refinement is a crucial step in protein structure prediction. Several prediction tools rely on the generation of very many low-resolution candidates and subsequent high-resolution refinement. Only few of the structures, however, are of sufficient quality to converge in the refinement step. Due to limited computer time it is therefore important to restrict the number of candidates and select only the few good ones. As the energy function used in the coarse-grained step is not very useful for recognizing good structures, we here discuss the use of structure profiles for this task. We show that the exact profile (derived from the native structure) is very reliable in choosing candidates with low cRMSD and TMscore to the native structure and clearly outperforms other methods such as filtering by energy or clustering. These profiles can also be predicted to good accuracy from the amino acid sequence. We therefore explore the use of sequence-derived profiles and demonstrate that for sufficiently high prediction accuracy this approach is also superior to the other methods of filtering and independent of the method used for coarse-grained structure generation [1].

[1] K. Wolff, M. Vendruscolo, and M. Porto, Proteins, 2009 in print, DOI 10.1002/prot.22533.