In our system, a universal oligo reference is defined as a single, labeled oligo that hybridizes to a common sequence on each microarray feature. In theory, this reference controls for differences in target DNA quality, spot morphology, and uneven hybridization, but not labeling or sequence-specific hybridization differences between transcripts. To evaluate the accuracy of this kind of reference relative to conventional microarray ratios, we compared data from hybridizations using conventional cDNA and oligo reference samples (table below) for yeast grown in four well-characterized media conditions, rich glucose (Glu), rich galactose (Gal), minimal raffinose (Raff), and minimal ethanol (EtOH). In these experiments, the calibrated oligo reference is a Cy3-labeled reverse complement of the universal priming sequence on the PCR products spotted on the arrays.

Conventional Universal Cy5 channel Glu cDNA Raff cDNA Cy3 channel Gal cDNA EtOH cDNA Cy5 channel Glu cDNA Gal cDNA Raff cDNA EtOH cDNA Cy3 channel Cy3 oligo Cy3 oligo Cy3 oligo Cy3 oligo

Statistical comparisons of the conventional microarray ratios and the reconstructed ratios (Gal:oligo/ Glu:oligo) for all spots on the array by two-tailed t tests and Wilcoxon rank sum analysis, produced results consistent with the hypothesis that the two sets of ratios were statistically equivalent. The graph below shows examples of Gal:Glu ratios from conventional hybridizations (blue) versus ratios reconstructed from oligo reference hybridizations (yellow) for two examples from four gene classes: highly induced in Gal (GAL1, GAL7), moderately induced in Gal (COX5A, QCR7), moderately induced in Glu (RPL3, RPL29), and equally expressed in both Glu and Gal (PHO88, STE5).

Below is a comparison of the information conveyed by conventional and universal oligo reference results for selected genes involved in galactose metabolism and mitochondrial function is shown in the two tables below. The first table shows results from cells grown in glucose versus galactose. Both systems detect the expected galactose induction with Gal:Glu >> 1 and Gal:oligo > Glu:oligo (P < 0.0001). The second table shows results from cells grown in raffinose vs. ethanol. Both systems demonstrate equal expression levels with Raff:EtOH ratios =~ 1 and Raff:oligo =~ EtOH:oligo (P =~ 0.24 for GAL genes, P =~ 0.4 for Mito genes). However, oligo reference results show that this is because GAL genes are not expressed while Mito genes are equally expressed 10-100 fold compared to GAL genes (Mito > GAL, P =~ 0.01 for both Raff:oligo and EtOH:oligo). All P values are the result of two-sided Wilcoxon rank sum tests. Thus, the ratios derived from the calibrated oligo reference hybridizations maintain absolute transcript abundance information far better than conventional microarrays.