Computational and Systems Biology Summer Institute
Iowa State University

We have been developing machine learning approaches to automated functional annotation of proteins. These methods have been used to discover potential errors in protein functional annotations, as well as for predicting protein subcellular localization, molecular function, etc . Work in progress is aimed at developing efficient, incrementally updatable predictive models for gene, protein, and genome annotation from large datasets. Of particular interest are methods for exploiting abstraction hierarchies and relationships among sequences to construct robust and accurate classifiers.