The research activity is focused on combinatorial optimization, probabilistic and message-passing algorithms, statistical mechanics, statistical physics of complex systems (disordered systems), out-of-equilibrium dynamics, analysis of algorithms and interdisciplinary applications of statistical physics (source and channel coding, game theoretical models of interacting agents, biological networks and neural computation).
The traditional approach to molecular biology has been an inherently local one, collecting and examining data on a single genes, proteins etc. In the last decades, biotechnological progress has made biological processes experimentally accessible on a genome-wide scale.
The challenge to utilize this data and turn it into an enhanced understanding of biological systems is, however, wide open. Statistical physics ideas might enter at least at two different levels. First, many of the underlying processes, like protein-protein or protein-DNA interactions are of physical nature, so data analysis should be guided by physical interaction models. Second, within the statistical physics of disordered systems new and highly efficient tools for the analysis of large optimization problems have emerged recently. The scope of our group is to adapt these methods to the needs in computational systems biology, and to apply them to the analysis of experimental biological data. Problem examples include the inference of gene-regulatory networks from gene-expression data, the analysis of metabolic networks, but also the development of robust algorithms for classical supervised and unsupervised learning problems as clustering and classification. In parallel we also aim at advances in the theoretical understanding of the performance and limits of the applied tools.