Prof. Dress headed the Department of Combinatorics and Geometry, which focus its studies on network interaction between proteins and metabolic products in healthy and diseased cells.

Large scale data are now available from various sources on the web such as the results of the numerous sequencing or structure determination projects, along with corresponding facilities for sequence and structure similarity searches. Our group has made extensive efforts in the area of protein sequence analyses. One major goal of our work is to develop a hybrid method which integrates information from different sources such as gene ontology (GO), protein function domains and the information content of protein sequences regarding the composition of amino acids or their physical and chemical properties. Such features can be used to predict protein structure classes, functional classes or protein subcellular locations, as well as protein-protein interaction. Our utilizes state-of-the-art methods, including Artificial Neural Network, Support Vector Machines or Nearest Neighbor algorithms. As an alternative to traditional machining learning approaches, we also employ concepts from graph theory in the context of proteome data analysis. Our primary goal is to develop more efficient algorithms based on determining maximal cliques in a graph in order to remove redundant biological sequences in large collections. While our work has already proven this approach to improve previous results, our ongoing research tries to apply dense subgraphs in predicting protein function based on the assumption functionally related proteins co-occur within one cluster in a protein-protein interaction network. Another branch of our work deals with small non-coding RNAs, most notably microRNAs, which play an important role in transcription regulation. In this context, we have developed computational approaches to predict targets of existing microRNAs.

The main themes of our scientific work include several important aspects of bioinformatics in human, animal, plant and microbial biology: information visualization, web services, databases, knowledge management tools for the above mentioned biological systems. The thrust of our themes is comparative biology, by comparing various species, we can infer unknown knowledge which cannot be discovered easily by experiments. The methodology consists of computation, mathematics, system theory and other engineering disciplines.

Experimental and computational aspects of the dynamics of eukaryotic cells