论文部分内容阅读
为了构建基因调控网络,提出一个基于生物先验数据融合构建非平稳动态贝叶斯网络结构的方法.该方法基于高斯混合网络模型,改变点过程和独立的能量函数.利用可逆跳跃马尔科夫蒙特卡罗抽样算法,把整个非平稳过程分解成若干平稳子片断,推断网络结构以及先验数据对网络的影响.在仿真和生物数据上测试该方法,结果显示该方法提高了网络重构的精度.
In order to construct a gene regulatory network, a new method based on bio-a priori data fusion is proposed to construct a non-stationary dynamic Bayesian network structure, which is based on the Gaussian mixture network model and changes the point process and independent energy function. By using the reversible jump Markov- Carlo sampling algorithm, the entire non-stationary process is decomposed into a number of stationary sub-fragments, infer the network structure and the impact of prior data on the network.Furthermore, the method is tested on the simulation and biological data, the results show that the method improves the accuracy of network reconstruction .