Deoxyribonucleic acid (DNA) microarray manufacturers have been through a rigorous market compression yielding only a few robust and trusted commercial platforms, each with signi cant market share. Agilent, Affymetrix, and Illumina remain the most ubiquitous and durable expression arrays, and have continued increasing density and quality. Many comparative analyses have shown the strengths and weaknesses of each platform, and have proposed methods to obtain the best crossplatform comparisons [1-12]. Several recent papers from the Microarray Quality Consortium (MAQC) [13-18] have shown that care in ribonucleic acid (RNA) preparation yields substantial bene ts in high correlation across different laboratories, platforms, and sample preparation methods. High-quality RNA remains one of the most important sources of variance, and although some platforms are more susceptible than others, the Achilles’ heel of commercial expression platforms is RNA integrity. Little can be done to rescue data from samples derived from degraded RNA, however normalization techniques help control for platform-speci c biases that are exacerbated by degraded samples. Unfortunately normalization methods are the second largest source of variance. An improper normalization will alter expression data even more than degraded RNA. Of the many normalization methods for expression data, each has its pros and cons. These can be leveraged in

order to obtain the most bene t from a normalization technique based on the type of array, the size of the experiment, and the hypotheses proposed by the experimenter.