一种基于分数阶微分方程模型的基因调控网络构建方法  

A Reconstruction Method of Gene Regulatory Network Based on Fractional Order Differential Equation

作　　者：季瑞瑞[1] 刘丁[1] 

机构地区：[1]西安理工大学自动化与信息工程学院,陕西西安710048

出　　处：《西安理工大学学报》2011年第2期127-132,共6页Journal of Xi'an University of Technology

基　　金：国家自然科学基金资助项目(61075044);西安理工大学学科联合研究基金资助项目(116-210905)

摘　　要：考虑到实际生物系统的非线性特性,提出了一种基于分数阶微分方程模型构建基因调控网络的新方法,采用模型预测数据与实际数据的逼近误差为目标函数,通过人工鱼群优化算法辨识分数阶微分方程模型的阶次和参数,并引入自适应步长,保留精英个体和增加种群多样性等策略提高算法的进化能力。对真实生物实验数据的结果表明,该方法能够较准确的辨识出模型参数,得到的分数阶微分方程模型与实际数据吻合程度较高。Considering the nonlinear characteristic of real biosystem,a novel reconstruction method based on fractional order differential equation is presented.The approximation error of predicted data to real data is considered as the objective function,artificial fish swarm algorithm is served as a tool to identify the power and parameters in the model.The evolvability of the optimization algorithm is enhanced through adaptive step,preserving excellent individuals and increasing the diversity of swarms.The experimental results on real life gene expression data show that this proposed method can infer gene regulatory model accurately and the fractional order differential model leads better data approximation.

关 键 词：基因调控网络 分数阶微分方程 人工鱼群算法 参数辨识 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]