Some statistical aspects of microarray analysis

Microarray experiments are generally designed to identify genes from different cell lines or treatment groups that are expressed at significantly different levels. The goal of the analysis of the experimental data is to generate an ordered list of genes that are likely to be differentially expressed between groups. Unfortunately, even the most carefully designed and executed microarray experiments have large amounts of variation in the observed experimental values. If the experiment has been designed with statistical modeling of the data as an important consideration, then a large proportion of the systematic variation in the data can be removed. With less variation in the data, a more accurate ordered list of differentially expressed genes can be prepared. In this paper we demonstrate the statistical modeling process for microarray data, and the result of removing systematic variation from microarray experimental data. This can lead to a more accurate ordered list of differentially expressed experimental genes, which helps improve the quality of the genetic or proteomic inquiry