Abstract: In order to study the impact of inner structure of biological systems and variations of correlative parameters on nuclear transcription fator-κappa B(NF-κB) signal transduction networks, it is vital to make sensitivity analysis of system parameters and to reduce mathematical model of NF-κB signal transduction networks, which includes 26 state variables and 64 parameters. Based on the analysis of mathematical model of the TNFα-Induced NF-κB signal transduction networks, the IκB Kinase complex（IKK）was chosen as step input signal of the system and NF-κB nuclear(NF-κBn) as measurable output signal. The direct differential method(DDM) was utilized to analyse sensitivity coefficients of oscillatory signal NF-κBn with respect to 64 parameters. Then, 9 parameters, which are less sensitive to system output signal, were removed form mathematic model of NF-κB signaling system so as to suitably reduce complexity of the system model. The simulation results show that output signal NF-κBn of the reduced model has much the same oscillatory characteristic as that of the former model. On the other hand, it also can be found that the rest output signals of both models are similar on the whole. Therefore, parameters sensitivity analysis and model reduction results can give new insights to analyse biological data, to build mathematical model and to design particular experiments.

Key words: sensitivity analysis, model reduction, NF-κB signal transduction networks, the direct differential method