Bioinformatics of the NF-kB-system 

The aim of the project is to explore and reconstruct the biological NF-kB transcription factor signal transduction system for better understanding. In particular we plan to map out the mechanisms that convert a chemical stimulus to a cell into a specific cellular response in the example of neural stem cells and adult neurons in mice.


Figure 1: Analysis on the integrated NF-kB protein-protein interaction and signaling transduction networks with edge bundling techniques and graph theory.

Due to the complexity of pathway interactions and large numbers of components involved in cell proliferation, cell differentiation, signal transduction, cellular rhythms, and cell-to-cell communication, it is quite difficult to intuitively understand the behavior of cellular networks. Recent experimental and computational progress yields networks of increased size and complexity that need to be examined carefully.

A common way to access the information in a network is through direct visualization, but this fails as it often only results in “fur balls” from which little insight can be gathered. Therefore we discover a new visualization approach, to highlight and communicate one particular piece of information about dynamic network structures. Moreover, we endeavor to find a lossless transformation of dynamic signaling networks into a compact, less redundant representation. We are investigating novel representations of networks, which reduce network complexity by explicitly representing re-occurring network motifs and dynamics, without any loss of information.

As a first result, we present a new approach to network visualization that tightly integrates network analysis methods and edge bundling techniques. The approach uses edge centrality measures to drive a force directed edge bundling method; this results in pictures that clearly show the most significant topological skeleton structures of the input network. We also introduce a new force-directed radial layout that shows group analysis of the k-cores: this results in pictures that show the important cohesive subgroup structures of the input network and their relationships (see figure 1).

The foundational modeling ideas and the relationship between network structure and system dynamics is a rapidly expanding frontier of this new science. Of particular interest is the understanding of the organization, complexity and dynamics of biological networks and how these are influenced by network evolution and functionality.

We aim at developing a theoretical framework for the function of specific signal integration modules in the local and global network context. Theories are in developing to address specific aspects of the network function, ranging from, for example, (1) graphical cell-based models to describe development and (2) network construction based on data and text mining to (3) quantitative models with implemented reaction kinetics and mass action relationships to describe information flow.

In order to reconstruct and analyze the NF-kB system we have begun to make use of data mining and information fusion by integrating databases with different contents: e.g. interacting partners and target genes. For the most important elements within the system we have generated experimental and database generated network systems that are regulated by the NF-kB transcription factor. The collected data and information is a combination of tightly interlinked complex systems at various levels of magnitudes.

Each network is constructed using a few basic mechanistic motifs/modules. The functionality of each network implicates participation of specific interacting proteins, where some proteins execute few, other many interactions. The networks operate in the dimension of time and network performance in agreement with cell requirements demands regulation of its activity, e.g. by feedback mechanisms to enable reliable cell fate decisions. Network regulation depends on additional interactions that are clearly visible in so called multidimensional networks.

The constructed models are examples of how the behavior of cells modeled by a Petri net can be simulated (see Figure 2).


Figure 2: Simulation of the NF-kB system in the Petri Net language

Ideas for laboratory experiments can be gained and tested by changing the basic Petri net. One approach is a theoretical knockout experiment. For this task, knocked out genes are modeled by deleting the corresponding transitions. A change of gene sequences influencing the catalytic efficiency can be modeled by changing speed parameters of the corresponding transition. Factors outside the cells that influence the signal molecules, the diffusion speed of the medium, and the availability of nutrients or even processes inside the cell can also be modeled.