EVOLUTION OF BIOLOGICAL REGULATION NETWORKS UNDER COMPLEX ENVIRONMENTAL CONSTRAINTS

This work investigates the influence of environmental inducers on the organization of cell regulation networks, using a connectionist approa ch. Protein interactions are modeled by an asymmetrical recurrent netw ork, the units of which take continuous values. In contrast to classic al models, we explicitly introduce a genome to encode the architecture of the system. This feature enables us to introduce an evolution mode l, in which a genetic algorithm that mimics the effects of evolution o n proteins mutual interactions is used. We assume an efficient system to respond to persistent environmental stimuli, independently of their amplitude. Results are presented that show a structuration of the net work with the emergence of specialized hierarchical structures. These structures seem to drive the system at the edge of chaos, so that it c an present adapted responses to significant environmental changes.