基于DNA微阵列数据分析的分级Bayes模型  

A Bayesian Hierarchical Model for DNA Microarray Data

作　　者：刘妍岩[1] 杨丹[2] 

机构地区：[1]武汉大学数学与统计学院,武汉430072 [2]武汉大学数学与统计学学院,武汉430072

出　　处：《应用数学学报》2009年第5期787-798,共12页Acta Mathematicae Applicatae Sinica

基　　金：国家自然科学基金(10771163)资助项目

摘　　要：如何分离出少量区别不同组织类型的特异性基因是DNA微阵列数据分析中的主要问题,特别是构建恰当的统计模型来刻画这些不同组织类型的DNA表达形式尤为重要.为此,基于基因DNA微阵列数据的特点,我们假定对数变换后的微阵列数据服从混合正态分布.我们采用分级Bayesian先验刻画不同基因的相关性,利用分级Bayesian方法构建模型,给出了刻画不同组织基因表达的差异的一个标准,用MCMC迭代计算该标准.模拟计算表明我们的模型具有较好的识别能力.In DNA microarray analysis, there is often interest in isolating a few genes that best discriminate between tissue types. In particular, it is critical to develop suitable models to explain the patterns of DNA expression for these different types of tissues. Toward this goal, we propose a methodology for the analysis of high-density oligonucleotide arrays. The log-transformed data are assumed to follow a mixture normal distribution based on the characteristic of gene itself. The variation in the data can reasonably be thought to arise from the effects of genes, tissue types, and their interactions. We introduce a hierarchical Bayesian priors for the parameters and propose a model selection criterion for identifving subsets of genes that show different expression levels between normal and tumor types. In addition, we develop Markov chain Monte Carlo algorithms for sampling from the posterior distribution of parameters and for computing criterion. The proposed methodology is evaluated via simulations studies.

关 键 词：BAYES推断 DNA 基因表达 分级先验分布 微阵列 混合正态 

分 类 号：O211.8[理学—概率论与数理统计]