Date |
December 22, 2005 |
Speaker |
Prof. Pierre Baldi, University of California, Irvine |
Title |
Prediction of protein structures on a genomic scale |
Abstract |
We will describe a full protein structure prediction pipeline that covers the spectrum of prediction approaches, from fold recognition to ab initio. The pipeline leverages evolutionary and statistical information using machine learning methods, such as recursive neural networks and kernel methods. The pipeline includes several components for the prediction of structural features (secondary structure, relative solvent accessibility, disulphide bridges), topological structure (contact maps, beta sheet architecture) and tertiary structure. Once the system is trained, predictions can be derived rapidly on a proteomic or protein-engineering scale. |
|