Persistence in metabolic nets.

In an attempt to improve the understanding of complex metabolic dynamic phenomena, we have analysed several 'metabolic networks', dynamical systems which, under a single formulation, take into account the activity of several catalytic dissipative structures, interconnected by substrate fluxes and regulatory signals. These metabolic networks exhibit a rich variety of self-organized dynamic patterns, with e.g., phase transitions emerging in the whole activity of each network. We apply Hurst's R/S analysis to several time series generated by these metabolic networks, and measure Hurst exponents H < 0.5 in most cases. This value of H, indicative of antipersistent processes, is detected at very high significance levels, estimated with detailed Monte Carlo simulations. These results show clearly the considered type of metabolic networks exhibit long-term memory phenomena.