The ill-posedness can be reduced by reduction of complexity either by shrinking the models to simplified network topology or by reducing the interaction between involved pathways to simplified mechanisms, such as boolean networks. In any case there will be payoffs by loosing biological features which are specific to the model. Hence, most applications tend to analyse the data using a set of models and C-1 decide according to a ranking of the respective model accuracies. A more generic drawback of fitting models to drugresponse surfaces arises when the MoA��s are not fully understood. Then the inherent issue arises that the model structure does not represent the biological mechanisms, which can lead to systematic errors in network reconstruction and model predictivity. Another approach to overcome the ill-posedness of models derived from combinatorial API-1 drug-response surfaces may be a systematic integration of multiple outputs of drug action into a network integrating drug descriptors, MoA and pharmacological data. Whereas the abovementioned approaches focus on specific pharmacological applications, combinatorial network reconstruction has been used to reconstruct generic signaling networks as well. Most of these approaches are based on evaluation of drug treatment on gene expression or protein phosphorylation profiles and the subsequent development of algorithms for reengineering of signaling networks. Combinatorial optimization algorithms are mostly being used in order to identify the relevant signaling networks out of a given set of pathway proteins. In principal, these networks have the potential to enable the identification of direct and specific drug targets or preferentially affected signaling pathways. Recently efficient, systematic and direct network reconstructions of induced phosphorylation of signaling proteins have been reported using combined stimulation and inhibition of cell cultures, where complex interaction networks have been reconstructed in detail from data describing combinatorial stimulation and inhibition of cells, using a highly multi-variate readout. Despite the tremendous improvement of understanding complex signaling networks and the interaction of the relevant pathways, drug effects mediated by yet unexpected cellular mechanisms, potentially as a secondary response on the primary drug action, may not sufficiently be assessed due to lacks in model structures. Hence novel unsupervised network reconstruction algorithms which are based on data obtained from broad-scale transcriptome and/or proteome profiling are needed as complementary method.