欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2013, Vol. 30 ›› Issue (6): 806-812.DOI: 10.7523/j.issn.2095-6134.2013.06.014

• 信息与电子科学 • 上一篇    下一篇

基于生物先验数据融合构建非平稳基因调控网络

倪小虹, 孙应飞   

  1. 中国科学院大学电子电气与通信工程学院, 北京 100049
  • 收稿日期:2012-09-12 修回日期:2013-03-26 发布日期:2013-11-15
  • 通讯作者: 倪小虹
  • 基金资助:

    Supported by the National Nature Science Foundation of China(60702035), Nature Science Foundation of Zhejiang Province(Y6090164)

Modelling non-stationary gene regulatory networks by combining microarrys with biological knowledge

NI Xiao-Hong, SUN Ying-Fei   

  1. School of Electronic, Electrical and Communication Engineering, University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2012-09-12 Revised:2013-03-26 Published:2013-11-15

摘要:

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.

关键词: 贝叶斯网络, 基因调控网络, 马尔科夫蒙特卡罗抽样, 改变点过程

Abstract:

In order to construct gene regulatory network, we propose a non-stationary dynamic Bayesian networks method that systematically integrates expression data with multiple sources of prior knowledge. Our method is based on Gaussian mixture Bayesian network model, change point process, and separate energy function of prior knowledge. Using an reversible jump Markov chain Monte Carlo sampling algorithm, we divide data into disjunct compartments, infer network structures, and measure the influence of the respective prior knowledge. Finally, we apply our approach to treat both synthetic data and biological data. The results show that the proposed method improves the network reconstruction accuracy.

Key words: Bayesian networks, gene regulatory networks, Markov chain Monte Carlo, change point process

中图分类号: