Gene-specific dye bias in microarray reference designs.

The most widely used microarray experiment design includes the use of a reference standard. Comparisons of gene expression between samples are facilitated because each sample is directly measured against the reference standard, using two fluorescent dyes. Numerous reports indicate that some genes incorporate the two commonly used dyes with different efficiencies, contributing to inaccurate data. However, it is widely assumed that these effects will not corrupt results if the reference standard is labeled with the same dye on each microarray. We demonstrate that this assumption is not reliable and that dye orientation can significantly influence measured changes in gene expression.