Modification and sensitivity of equilibria in partially characterized genetic networks

Abstract

Engineering intracellular biological control systems, whether to perform artificial functions or to modify existing behavior, requires accurate characterization of genetic networks. Slight errors in characterization lead to significant changes in predicted behavior, and consequently to design errors that are difficult to debug. Metabolic engineering provides an empirical solution to this problem by varying many system parameters simultaneously to optimize the steady state level of a particular chemical species. Such approaches, though effective, require significant customization for the systems being studied. Here I propose and investigate, in silico via generation and analysis of random genetic networks, a more general approach in which a small number of modifications, selected empirically, are made to genetic networks. We found that feedback loop based modifications were most effective, and that responses to modifications depended significantly on network structure. Our work serves to advance a potential technique for efficient empirical design and modification of genetic networks.