Simulation of Circadian Rhythms Using Grid Computing

The Drosophila circadian oscillator shows a robust property to environmental changes, internal variations, and fluctuations, where two interlocked negative feedback loops play a major role for oscillation [1, 2]. The PER-TIM loop is activated by dCLK-CYC and repressed by PER-TIM, whereas the dCLK loop is repressed by dCLK-CYC and activated by PER-TIM. Not only the molecular architecture, but also the parameters provide a robust property to the oscillator. To routinely construct the dynamic model of a biological system, we propose a novel and powerful strategy that consists of forward and reverse engineering. Forward engineering produces a mathematical model with kinetics-related parameters directly from biochemical networks. However, since it is hard to obtain all the real kinetic values in vivo experimentally, the use of reverse engineering explores the kinetic parameters of the dynamic model. In order to estimate the values of the kinetic parameters that reproduce the dynamic features of the oscillator, we employed a random search (RS) and genetic algorithms (GAs). However, since there are many candidates of the kinetic parameter vectors that reproduce the similar dynamic behaviors, it is hard to obtain a unique solution. In order to select a unique solution out of many candidates, we carried out the sensitivity analysis by calculating the changes in the period or in the amplitudes with respect to the variations in each parameter. Since we needed extensive calculations for performing RS and GA, we used grid computing.