In the absence of a standard format for storing microarray information, people— oftentimes within the same institution—find it difficult to use the millions of data points accumulating in the several microarray repositories extant today. Since its inception in 1999, the Microarray Gene Expression Data Society (MGED) has been working to solve this problem, and the group put forth its latest attempt, MAGE-ML, in the 23 August issue of Genome Biology, (a publication of The Scientist's partner, BioMed Central).

MAGE-ML (Microarray Gene Expression Mark-up Language) follows on the heels of an earlier MGED-devised standard, MIAME (Minimum Information About a Microarray Experiment), which defined for users what information is needed to facilitate data sharing. MAGE-ML provides a framework for moving and storing microarray data sets that incorporates the principles set down in MIAME, bringing the field one step closer to having universally usable public repositories.

The difficulty with microarray experiments goes beyond simple data handling. A single experiment can generate upwards of several million pieces of data, but without other information—biological material, experimental design, array composition and design, hybridization conditions, and so forth—the data is largely uninterpretable. "With gene expression, context is everything," noted Michael Ashburner, of the European Bioinformatics Institute, in a 2000 commentary published after MGED's first meeting.

In order for repositories of expression arrays to be useful, they must contain lots of bits of information and contain them in such a way that the information can be queried and transmitted along with the data. "MAGE helps get that information into a database, and helps you figure out, if you dump it from a database, what it all meant," MGED board member Paul Spellman, from the University of California, Berkeley, told The Scientist.

According to Spellman, the working group, which included both academic and industry researchers, tried to anticipate new kinds of arrays and ways they might be used in the future. "It was designed to be as flexible as possible in terms of supporting anything we could think of. It's so flexible that with just one change in the model, it can support data from tissue microarrays," he added.

MGED hopes that eventually journals will require researchers to submit microarray data from their publications into repositories, as happened with DNA sequences. However, Spellman thinks that with microarrays, this could take a while. "It happened pretty quickly with sequences, [but] MIAME suggestions to authors and journals are pretty radical in scope. Journals are leery because it represents a major increase in complexity," he said.

Mike Ocks, Director of Bioinformatics at Fox Chase Cancer Center in Philadelphia sees MAGE-ML as a positive step, as it takes some of the pain out of sharing information. "It will give us the ability to look at a lot of other data with other methods of analysis to see if we can see something new. New methods come along every few months. Wouldn't it be nice to go back [to old data], but it's very hard to get the data out. This should speed it up," he said.