Boolean networks are widely used in systems and computational biology to model regulatory relations among genes. In the past, investigations with respect to the stability of such networks have been performed. A network is considered stable, if a perturbation of a node will vanish after a certain number of iterations. In particular, in random Boolean networks an order coefficient lambda has been introduced. If lambda is smaller than 1, a network is called stable or ordered, i.e., an error will not propagate much. On the other hand, if lambda is larger than 1, a network is called chaotic.

The aim of this semester thesis is to investigate the stability of such regulatory Boolean networks. The focus lies on two regulatory networks introduced in the literature by Covert et al. and Feist et al.. These networks  have a layered feed-forward structure, where only the status of the lowest layer, the "output" layer, is of interest. The effects of small disturbances, e.g., mutations, of the input layer on the output of the network is to be studied.

First the networks are to be analyzed with respect to topology, in- and out-degree distribution and other statistics. Moreover the expected average sensitivity of the functions is to be determined, as its value corresponds to lambda. Using these values, one can predict the behavior of the error propagation in a randomized networks.

By performing a series of simulations for the networks found in literature as well as randomly generated networks the actual error propagation is to be investigated and compared to the theoretical results.