The methods we develop advance the understanding of gene function by analysis of complex, heterogeneous experimental data including data from imaging. Computer vision methods combined with statistical models help to find functional modules in Drosophila development by their spatial and temporal co-expression patterns. Novel tree-models elucidate regulatory mechanisms in the development of the lymphoid system (see a recent publication tagged as at BMC Immunology ).
We work on theory and application of classical bioinformatics tools such as Hidden Markov Models and mixture models to novel data with a particular focus on semi-supervised learning and flexible, robust models with minimal number of parameters (CSI).
Screening experiments as well as the design of DNA microarrays can be optimized by advanced statistical approaches such as group testing. A Dagstuhl seminar in the summer of 2008 will give an overview of applications in the life sciences.
Teaching algorithms and methods benefits greatly from electronic media. We develop animation systems for graph algorithms and clustering algorithms; CATBox is a forthcoming Springer book. Learners can concentrate on tackling exciting bioinformatics problems with our Hidden Markov Model library.