基于扩展卡尔曼滤波的信号转导网络参数估计

Parameters Estimation of Signal Transduction Networks Based on Extended Kalman Filter

针对细胞信号转导网络的数学模型结构复杂、强非线性以及实验测量数据存在极大不确定性的特点,采用扩展卡尔曼滤波技术,将模型参数作为随机过程的状态估计量,构造相应的卡尔曼滤波方程进行参数估计.以肿瘤坏死因子诱导的核转录因子κB信号转导网络为例,对系统模型的未知参数进行辨识.结果表明该方法能够从噪声数据中提取有效信号,从而有效地识别参数,为解决复杂信号转导通路参数辨识中的不确定性问题提供了可靠的理论方法.

The mathematical model of cell signal transduetion networks is highly complex and strongly nonlinear since it includes a lot of dynamic parameters and reaction species. Due to the impact of the noise, the experimental data are incomplete and uncertain. The extended Kalman filter method is pre- sented for parameters identification of signal transduetion networks model. The unknown parameters are improved iteratively as the state vector of a random process so that the Kalman filter equations can be formulated to perform parameter identification. Using TNFα induced NF-κB signal transduetion network as an example, this method was applied to estimate the unknown parameters of the model. The simulation results demonstrate that the algorithm can well estimate the unknown parameters under the disturbing of the noise and it provides an efficient theoretical tool for solving the parameters＇ uncertainty effects of biological pathways.