A personalized microRNA microarray normalization method using a logistic regression model
A personalized microRNA microarray normalization method using a logistic regression model
复制标题
DOI：
10.1093/bioinformatics/btp655
复制
发表时间：
2010-01-15
期刊：
影响因子：
5.8
通讯作者：
Xi, Yaguang
中科院分区：
文献类型：
--
作者：
Wang, Bin;Wang, Xiao-Feng;Xi, Yaguang
MicroRNA（miRNA）是一组新发现的非编码小RNA分子。它的显著作用导致了许多关键的生物学事件，包括细胞增殖、凋亡发展以及肿瘤发生。高维基因组发现平台（e. G.微阵列）已经被用来通过分析它们的表达谱来评估miRNA的重要作用。然而，由于miRNA的数量少，缺乏已知的内源性对照，传统的mRNA谱分析标准化方法不能提供合适的miRNA分析解决方案。结果：采用基于锁核酸（Locked nucleic acid，LNA）的miRNA芯片，对不同处理条件下的结直肠癌细胞株进行miRNA谱分析。使用基于Taqman的定量实时聚合酶链反应（qRT-PCR）miRNA测定通过一组miRNA预评估总体miRNA谱的表达模式。基于qRT-PCR结果建立逻辑回归模型，然后将其应用于miRNA阵列数据的归一化。对从标准化列表中选择的20种另外的miRNA的表达水平进行后验证。与其他常用的标准化方法相比，逻辑回归模型有效地校准了阵列间的方差，提高了miRNA微阵列发现的准确性。
Motivation: MicroRNA (miRNA) is a set of newly discovered non-coding small RNA molecules. Its significant effects have contributed to a number of critical biological events including cell proliferation, apoptosis development, as well as tumorigenesis. High-dimensional genomic discovery platforms (e. g. microarray) have been employed to evaluate the important roles of miRNAs by analyzing their expression profiling. However, because of the small total number of miRNAs and the absence of well-known endogenous controls, the traditional normalization methods for messenger RNA (mRNA) profiling analysis could not offer a suitable solution for miRNA analysis. The need for the establishment of new adaptive methods has come to the forefront.Results: Locked nucleic acid (LNA)-based miRNA array was employed to profile miRNAs using colorectal cancer cell lines under different treatments. The expression pattern of overall miRNA profiling was pre-evaluated by a panel of miRNAs using Taqman-based quantitative real-time polymerase chain reaction (qRT-PCR) miRNA assays. A logistic regression model was built based on qRT-PCR results and then applied to the normalization of miRNA array data. The expression levels of 20 additional miRNAs selected from the normalized list were post-validated. Compared with other popularly used normalization methods, the logistic regression model efficiently calibrates the variance across arrays and improves miRNA microarray discovery accuracy.