Hierarchical Probabilistic Interaction Modeling for Multiple Gene Expression Replicates
Hierarchical Probabilistic Interaction Modeling for Multiple Gene Expression Replicates
复制标题
DOI：
10.1109/tcbb.2014.2299804
复制
发表时间：
2014-03
期刊：
影响因子：
--
通讯作者：
Kristopher L. Patton;D. J. John;J. Norris;Daniel R. Lewis;G. Muday
中科院分区：
文献类型：
--
作者：
Kristopher L. Patton;D. J. John;J. Norris;Daniel R. Lewis;G. Muday
微阵列技术允许在少数时间点收集数百个基因的基因表达时程数据的多次重复。基于时间进程基因表达数据发展关于基因转录网络的假设是一个重要且非常具有挑战性的问题。在许多情况下，存在相似性，这表明重复之间存在层次结构。本文提出了基于多层次复制的网络特征的后验概率。通过贝叶斯推理，结合Metropolis-Hastings算法和模型平均，分层多重复算法应用于七组模拟数据和一组拟南芥基因表达数据。模拟数据的模型表明，具有至少中等信号偏相关性的基因对的后验概率较高。对于拟南芥模型，许多最高后验概率边缘与文献一致。
Microarray technology allows for the collection of multiple replicates of gene expression time course data for hundreds of genes at a handful of time points. Developing hypotheses about a gene transcriptional network, based on time course gene expression data is an important and very challenging problem. In many situations there are similarities which suggest a hierarchical structure between the replicates. This paper develops posterior probabilities for network features based on multiple hierarchical replications. Through Bayesian inference, in conjunction with the Metropolis-Hastings algorithm and model averaging, a hierarchical multiple replicate algorithm is applied to seven sets of simulated data and to a set of Arabidopsis thaliana gene expression data. The models of the simulated data suggest high posterior probabilities for pairs of genes which have at least moderate signal partial correlation. For the Arabidopsis model, many of the highest posterior probability edges agree with the literature.