Center for Statistical Sciences and Department of Community Health, Brown University, RI 02912, USA.
Stat Methods Med Res. 2009 Dec;18(6):533-41. doi: 10.1177/0962280209351924.
Microarrays have become an indispensable tool in biomedical research. This powerful technology not only makes it possible to quantify a large number of nucleic acid molecules simultaneously, but also produces data with many sources of noise. A number of preprocessing steps are therefore necessary to convert the raw data, usually in the form of hybridisation images, to measures of biological meaning that can be used in further statistical analysis. Preprocessing of oligonucleotide arrays includes image processing, background adjustment, data normalisation/transformation and sometimes summarisation when multiple probes are used to target one genomic unit. In this article, we review the issues encountered in each preprocessing step and introduce the statistical models and methods in preprocessing.
微阵列已成为生物医学研究中不可或缺的工具。这项强大的技术不仅使同时定量大量核酸分子成为可能,而且还产生了许多噪声源的数据。因此,需要进行一些预处理步骤,以便将原始数据(通常以杂交图像的形式)转换为可用于进一步统计分析的生物意义的度量。寡核苷酸阵列的预处理包括图像处理、背景调整、数据归一化/转换,以及当多个探针用于靶向一个基因组单元时的汇总。在本文中,我们回顾了每个预处理步骤中遇到的问题,并介绍了预处理中的统计模型和方法。