A comprehensive analysis of the effect of microarray data preprocessing methods on differentially expressed transcript selection

Normalization techniques have a profound influence on microarray expression data. Recently, in an attempt to create a robust consensus across multiple expression platforms, several well-meaning groups have proposed methods that strongly weight cross-platform concordance. However, when normalization techniques affect downstream analysis on the same platform in fact, when they are applied to the same data are we getting ahead of ourselves by trying to achieve cross-platform concordance without satisfactorily addressing this problem first? The authors have investigated the effects of normalization techniques on the same platform and same data set. They present a thorough report on arguably the most important output of expression analysis: the list of genes shown to be differentially expressed across conditions, which is a usual step following expression data generation. This comprehensive approach provides the reader with a nonbiased view of the effect of normalization on detection of differential expression.