Boston

A grassroots collection of biologists plans to persuade scientific journals to enforce minimum standards for the publication of all experiments involving microarrays.

Genetic diversity: microarray experimental results are affected by a multitude of variables. Credit: ZmBD/DAVID GALBRAITH/UNIV. ARIZONA

The arrays, which monitor the expression of thousands of genes simultaneously, have proved to be helpful in the discovery of genes that influence such diverse processes as cancer and embryonic development. But whereas DNA sequences are always in the same format, microarray experiments are difficult to compare with one another, because of the multitude of variables that affect their outputs (see Nature 410, 851; 2001 and Nature 410, 860–861; 2001).

In an effort to remedy this, the Microarray Gene Expression Database (MGED) Group first devised a common set of standards — called minimum information about a microarray experiment (MIAME) — in 1999.

But although many papers now refer to the MIAME standards, compliance is uneven, says Catherine Ball, curator of the Stanford Microarray Database. “At the moment, it's terrible,” she says. “I often find I can't identify genes or that people have used home-grown software they don't describe.”

“Generally, we want the same thing you have with published DNA sequences,” says Alvis Brazma, head of microarray informatics at the European Bioinformatics Institute in Cambridge, UK. This means that all primary research data should be submitted to a repository, he says. The standards would also require the publication of enough experimental detail to allow other scientists to compare different data sets.

In early March, the MGED steering committee, of which Ball and Brazma are members, plans to release a check-list based on MIAME for use by authors, editors and referees. The list will be designed to make it easy for journals to request, and ultimately require, MIAME compliance.

The final check-list is likely to ask for details of sample preparation and data processing, as well as numerical read-outs of the intensity of the array's red and green spots, which indicate the level of gene expression.

Ball says that the check-list won't require authors to reveal every gene on an array, as this could allow others to scoop experiments in progress. “No one would argue that you should publish data prematurely,” she says. “You can strip out the gene identity of things you are not publishing.”

Journal editors are enthusiastic about the proposed check-list, but say that they will consult the microarray researchers before requiring authors to comply with it. “We'd consider insisting if the community showed broad consensus,” says Nature's editor, Philip Campbell.

Laurie Goodman, executive editor of Genome Research, published by Cold Spring Harbor Laboratory, says that she would be cautious about requirements that could rule out important papers, but would consider adopting minimum standards if she received a strong indication that microarray users support them.

→ http://www.mged.org