Microarray standard data set and figures of merit for comparing data processing methods and experiment designs: Supplementary

SYSTEMS AND METHODS Experiment Design and Rationale The data set was designed to provide a range of difficulty for pattern matching and detection of expression changes. mRNA samples A through T from 20 different cell lines and tissues listed in Table S1 were combined in different ways to provide condition pairs with both subtle and moderate expected expression differences. Figure S1 summarizes the sample pairings for the two-color hybridizations. Each connecting line segment represents a pair of arrays where each array is hybridized with the two samples but the assignment of fluorescent labels is reversed. This fluor reversal strategy allows removal of most of the biases caused by dye-specific interactions during the process. Each fluor-reversed-pair (FRP) of arrays in which the samples in Cy3 and Cy5 were non-identical was repeated to provide the redundancy crucial to the proposed Type 2 figure of merit evaluation method. Thus each connector in Figure S1 actually represents four microarray hybridization experiments. The overall experiment design can be thought of as a ring with spokes and chords. The ring design allows constructing arbitrary pairwise comparisons in silico and has some advantages over comparing each condition with the same reference sample. The chords allow comparing direct