GNN - Genome News Network  
  Home | About | Topics
Standards proposed for gene expression data
Edward R. Winstead


Versatile and popular, DNA microarrays have been used to diagnose disease, compare genomes, and classify new types of cancer. In just the last few years, an enormous amount of data has been published. Citing the diversity of the information, a group of researchers has proposed ideas for standardizing gene expression data—the recording and reporting of when thousands of genes are turned on or off. Their goal is to create standards that make expression data easier to share, interpret, and verify by researchers worldwide.

Microarray from TIGR

The proposal is the work of a self-described "grass-roots movement" called the Microarray Gene Expression Database (MGED) group, which includes academic, government, and industry researchers with expertise in microarrays and bioinformatics. The group was formed in 1999 and has held three general conferences on the topic of establishing voluntary standards.

"The goal of our proposal is not to impose specific solutions upon the community, but instead to establish a community-wide understanding of the optimal infrastructure for the sharing of microarray data," the researchers write in a commentary in Nature Genetics. They have set up an e-mail discussion group to promote an exchange of views.

Standards are needed, they say, because expression data are meaningful only when interpreted in the context of the experiment. To draw conclusions about expression profiles of breast cancer tissue, for example, one needs to know the genetic background of the donor and the donor's stage of disease.

Even with detailed information about how the experiments were done, comparing expression data derived from different experiments is difficult. Microarrays, or gene chips, do not currently measure gene expression levels in objective units; most measure relative changes in gene expression. Furthermore, many types of DNA microarrays are publicly available, and laboratories build their own custom tools—all of which contributes to the non-uniformity of data.

In developing the proposal, the researchers write, "we have sought to find a compromise between placing a burden on data producers to annotate experiments in elaborate detail and ensuring that data are annotated in enough detail to be useful to the general research community." The proposal goes into detail in six parts of the experimental process: experimental design, array design, samples, hybridizations, measurements, and normalization controls.

The adoption of standards, they believe, will facilitate the creation of microarray databases. The Microarray Gene Expression Database group has a fourth general meeting planned for February 2002 in Boston, Massachusetts. For more information about the group visit

. . .

Brazma, A. et al. Minimum information about a microarray experiment (MIAME)—toward standards for microarray data. Nat Genet 29, 365-371 (December 2001).

Back to GNN Home Page