Experimental Design & Analysis of Protein Array Data: Applying Methods from cDNA Arrays.

It appears that a number of recent manuscripts using protein microarray technology are using equivalent analysis procedures that the gene-expression microarray community implemented in their infancy. That is, utilizing a classic reference design such that the ratio of the sample of interest to a reference sample is the response of interest and assessing fold change to determine differential expression. For example, recent publications have concluded that proteins with a fold change less than 0.7 or greater than 1.3 demonstrate significant down- or up-regulated differential expression, respectively. However, fold change is an unreliable measure of differential expression and statistical models that distinguish true signal from random noise should be utilized instead of fold changes. Over the last half decade a tremendous amount of research has been devoted to gene-expression microarrays to vastly improve on the areas of experimental design, normalization and statistical analyses to assess differential expression and classification and these methods are directly applicable to protein microarray technology. Thus, the objective is to review the statistical methodology that has been developed for two-color cDNA arrays that is directly applicable to protein arrays. Examples are provided from a mantle-cell lymphoma protein-array experiment.