Zuverlässigkeit und Evolvierbarkeit genetischer Regulationsnetze

Living organisms are remarkably robust despite fluctuating concentrations of functional molecules in the cell and changing environmental conditions. In the biological literature, the question how organisms cope with this stochasticity has been investigated in theory and experiment in specific organisms. To identify and understand general mechanisms that facilitate reliable dynamical behavior, computer modelling can be useful to investigate specific effects in isolation. In this thesis, the effect of the topological structure of transcriptional regulation networks on the reliability of the resulting dynamics is investigated in simple dynamical models. The activity of genes and proteins is modeled by discrete values. An extension of this discrete dynamical model to continuous time is used and molecular fluctuations are implemented by random delays of signals. Reliability of the dynamics is defined as ordered dynamical behavior despite these fluctuations. Using this criterion, simple systems of interacting genes as well as the model organism budding yeast S. cerevisiae are investigated. The reliability of the cell-cycle regulation is assessed and simple mechanisms of the regulational organization are identified which lead to the robust dynamical behavior. Further, the recently discovered feature of biological networks to display a non-random distribution of interaction patterns among triads of nodes, the 'motif distribution,' is investigated. In a simple evolutionary model using a suitable selection criterion, dynamically robust networks are produced. However, these networks do not display the expected motif distributions. This points to an ability of the evolution model to create reliable dynamics without significant changes of the network structure. To further explore this, it is investigated how easily reliable networks emerge in an evolution process sing different models: First, the reliability of all dynamical attractors of networks shall be accomplished. An astonishing evolvability towards reliable dynamics is observed. Second, a specific 'functional' attractor is defined that has to be reproduced reliably to reach the goal of the evolution. Most such evolution processes successfully finish in this criterion. These results indicate that dynamical reliability is an evolvable property of regulatory systems