Leming Shi of the US Food and Drug Administration (FDA) in Rockville, Maryland, is unabashed in his affection for microarrays. But he was disconcerted by the recent publication of several papers challenging the reliability of gene-expression microarray experiments. One article, for example, reported such disagreement that an analysis of 185 genes using three different technologies revealed concordant readings for only four transcripts (P. K. Tan et al. Nucl. Acid Res. 31, 5676–5684; 2003).

Leming Shi hopes to improve the reproducibility and comparability of microarray work.

Subsequently, Shi and his colleagues found that an alternative analytical approach greatly improved cross-platform concordance for these data sets (L. Shi et al. BMC Bioinformatics 6 (Suppl. 2), S12; 2005). But the lingering climate of uncertainty, and concerns about the potentially serious implications for the use of microarray data in the FDA drug-approval process, led them to launch the MicroArray Quality Control (MAQC) project. The MAQC brought together research leaders from government, academia and industry to establish tightly controlled 'gold standard' comparisons of microarray systems. They began by identifying commercially available, trustworthy 'standard' RNA samples. But this was just the start. “The MAQC's main goal is to generate a vast reference data set,” says Shi. “We have conducted more than 1,000 array hybridizations with these reference samples, plus we're using three alternative technologies and we requested that each system be evaluated at three testing sites.” The MAQC recently completed a review of its final data, and will present its findings in a series of articles to be published in Nature Biotechnology next month.

In the meantime, many in the field are awaiting the outcome of a complementary initative: the External RNA Controls Consortium (ERCC). This evolved from a 2003 meeting headed by the US National Institute of Standards and Technology in Gaithersburg, Maryland. It aims to identify and help make commercially available a collection of reliable RNA 'spike-in' controls, which can be included in any microarray experiment to assess variables such as labelling and hybridization efficiency. Participation has grown rapidly, and ERCC leader Janet Warrington, who is a vice-president at Affymetrix in Santa Clara, California, finds the early progress promising. “A number of organizations that were already using their own controls have donated these — no strings attached — for testing,” she says. “So we have a collection of 100 to 150 controls that will be tested across platforms and we have eight sites that have volunteered to carry out the testing.”

Both projects have benefited from collaborative environments that have allowed even direct competitors to work together towards a shared goal. “We all share the belief that if we're successful, we'll expand the marketplace for everyone,” says Warrington.

M.E.