基于不同标准化假设分析癌基因组甲基化谱  

Analyses of DNA Methylation Data Based on Different Normalization Assumptions

作　　者：王栋[1,2] 吴瑞红[2] 程立新[2] 李朋飞[2] 王明月[2] 李宾[2] 王晨光[2] 郭政[1,2] 

机构地区：[1]电子科技大学生命学院,成都610054 [2]哈尔滨医科大学生物信息科学与技术学院,哈尔滨150086

出　　处：《中国生物医学工程学报》2011年第5期706-711,共6页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金重大研究计划(91029717);国家自然科学基金(81071646);黑龙江省自然科学基金(QC2010012);黑龙江省教育厅科学技术研究项目(11541156)

摘　　要：在处理甲基化谱时,通常采用标准化方法,强制所有样本的信号值服从同一种分布或至少具有相同的中值。然而,有研究者认为这些标准化方法可能会移除甲基化谱中的生物学信号,有必要全面评估不同的处理方法对后续分析的影响。通过配对秩和检验,分析关于4种癌型的8套甲基化谱。结果显示,并未发现癌与正常样本中的原始信号值的中值存在显著差异(P>0.05);采用分位数标准化数据后可以多筛选出12%～28%的差异甲基化基因,而且额外找到的差异甲基化基因的改变方向可以在关于同种癌型的独立数据集中显著一致地呈现(P<0.001),表明它们是有效的生物学信号。但是,在经过分位数标准化后筛选出的差异甲基化基因中,约有1%基因的甲基化状态与原始信号的改变方向相反。建议采用分位数标准化方法处理甲基化谱数据,但需要去除改变方向不一致的基因。Normalizations for methylation data usually force all arrays to have the same probe intensity distribution regardless of the biological groups of samples.However,some researchers have suggested that most of current normalization methods may remove real biological signals of DNA methylation data.In this paper,using eight datasets of methylation for four cancer types,we showed that the medians of methylation signals in normal samples were not significantly different（P0.05）from those in cancer samples,and more differential methylation（DM） genes（by 12%~28%） could be found using quantile method to normalize data compared with using non-normalized data.We found that the DM genes solely selected by using quantile method in one dataset for one cancer have significantly consistent methylation states（P0.001） in another independent dataset for the same cancer,indicating that these extra DM genes are effective biological signals.However,for one cancer,the methylation states of about 1% of DM genes selected in quantile-normalized data are opposite to those in the non-normalized data.Hence,we must filter out these false DM genes.

关 键 词：DNA甲基化 微阵列 数据标准化 癌基因组 差异甲基化 

分 类 号：R318[医药卫生—生物医学工程]