DNA microarrays from the leading commercial sources are more reliably consistent than previously suspected, according to a new study coordinated by the U.S. Food and Drug Administration (FDA).
The MicroArray Quality Control (MAQC) project, directed by Leming Shi, an FDA scientist at the National Center for Toxicological Research in Jefferson, Arkansas, compared seven different DNA microarray platforms with regard to their reliability from experiment to experiment as well as the optimal approaches for analyzing resulting data.
The study included microarrays from five major vendors: Affymetrix, Agilent, Applied Biosystems, GE Healthcare, and Illumina, and is published online in Nature BioTechnology.
Shi didn’t agree with several previous studies that reported inconsistencies amongst various microarray gene expression studies. So in 2004-05, Shi set about designing a better study, ultimately involving a total of 137 scientists from 51 scientific organizations. Says Illumina’s Shawn Baker, scientific product manager, gene expression: “He rallied all the major microarray participants together.”
After a couple of pilot stages, during which Baker says one vendor decided to opt out, the five aforementioned vendors plus Eppendorf continued onto the major part of the study. A seventh system consisted of National Cancer Institute arrays using oligonucleotides from Operon.
The MAQC study used two RNA samples – a Universal Human Reference RNA from Stratagene, and a human brain reference sample from Ambion. Four test samples were prepared – the two pure RNA samples, as well as two mixtures in 1:3 and 3:1 ratios.
For each microarray platform, each of the four samples was tested in five replicates, or 20 microarrays. Furthermore, each set of 20 microarrays was compared between three different sites, including the major manufacturing site for each array. So a total of 60 arrays were tested per vendor platform. The results were validated using alternative technologies such as TaqMan PCR assays.
Concordant Results
The experimental data were gathered in late 2005, followed by analysis and manuscript preparation and review. Baker professed to be delighted with the results of the study. “These various platforms were able to get very concordant results, in contrast to a few but not all previous studies. Technical performances were very similar across all platforms.” Not surprisingly, Baker added, “Illumina tended to perform at the high end of technical performance.”
The first objective, says Baker, was “to determine how concordant results were across platforms.” In other words, how similar were the differences between the two RNA samples when comparing the various platforms?
“Technical performance was very similar across the platforms,” says Baker. “Illumina was always at the top end. We shine in value—the cost of our arrays is substantially less than all other providers. All other things being equal, that’s going to give tremendous power to people using our platform. With the same budget, they can dramatically increase the number of samples, replicates, and biological conditions to increase the power of the study.”
Baker says despite the commercial competition, the spirit among the participants was “very congenial.” He added: “I wasn’t sure what to expect given the commercial pressures. I think it’s a tribute to the participants and to Leming and his leadership.”
The FDA considers pharmacogenomics and toxicogenomics important staples of the personalized medicine revolution, and recognizes the critical importance of microarray studies in furthering those fields. The results of the MAQC Consortium should reassure scientists, clinicians, and the FDA, that most major commercial platforms can be selected with confidence. Moreover, the MAQC dataset offers scientists a highly validated reference for future work.
-----
Featured Paper: MAQC Consortium: “The Microarray Quality Control (MAQC) project: shows interplatform reproducibility of gene expression measurements.” Nature Biotechnology AOP.