A novel 4-valued discrete network for the analysis of signalling events integrating protein interaction and gene expression data applied on hematopoietic diefferentiation

Signaltransduction is an important process, that is crucial for the interplay of cells in multicellular organisms and controls among other things cellular proliferation and cellular differentiation. The complexity of signalling networks and the absence of comprehensive and accurate quantitative data make a precise mathematical description of such networks difficult. This work deals with the questions, if and how predications about signalling processes for specific cellular models can be made with the aid of interaction databases and gene expression profiles. Starting from molecular interaction-databases complex networks are generated comprising all known cellular signalling paths. But, not all of them play a role in a specific cellular model. The specific adaptation of the networks to a specific cellular system is achieved by additional integration of data obtained from gene expression profiles. Thereby, 1. the information about expression or non-expression of a gene and 2. the change of gene expression in time (increase, decrease or unchanged) is used. 1. Information about the expression is included in the calculations, by removing unexpressed elements from the network and thus changing its topology. 2. Chronological changes in expression are integrated in the estimations by a four-valued, discrete network containing statistical elements, that was developed in this work. This network is able to deal with contradicting data and pays attention to the importance of feedback loops for differentiation processes. This enables an extraction of subnetworks having a dynamic that approximates the measured data best. Hematopoietic processes, playing among other things a role in development of leukaemias, were examined with a sensitivity analysis. The analysis revealed central molecules of EpoR induced hematopoesis, like AKT, PIP3, Pi3K, Lyn, ERK