London

Biotechnology company Affymetrix is retreating from a plan that would have made it tougher for biologists to use open-source computer software to analyse results from its ubiquitous microarray chips.

The firm, based in Santa Clara, California, had planned to upgrade its microarrays, which are used by thousands of molecular biologists to detect gene activity. It hoped that the upgrade would have made the output files from its chips smaller and faster to process.

But researchers energetically protested that the new files would be incompatible with open-source software that has been developed by academics to work with the company's microarrays. Such software is free for everyone to use.

On 2 July, Affymetrix said that it would change course to ensure that its upgraded technology would not render the free software obsolete.

Microarray chips are covered in thousands of strands of DNA, each corresponding to a known gene and marked with a fluorescent molecular tag. Researchers add a sample of biological material to the chip, and the DNA in the sample bonds to the sequences on the chip. The chip outputs its results — a record of which dots are glowing, and how brightly — as a computer file. At present, these files can be opened on almost any computer.

Affymetrix announced late last year that it would upgrade its microarrays to produce data as smaller files in a proprietary format. The change would have made it easier to modify formats in the future without disrupting the analysis software, says David Kulp, a bioinformaticist at the company.

But the move would have placed the microarray data files off-limits to the developers of open-source software. It would also have rendered useless many existing tools for interpreting the data, including a popular one called Bioconductor.

Many molecular biologists are happy with the simple file format produced by Bioconductor. “The file is readable by a human,” says James MacDonald, who uses Bioconductor to analyse output from chips at the University of Michigan's microarray facility at Ann Arbor.

Researchers' worries were increased by the fact that US law now makes it difficult to write open-source software to interface with proprietary formats. On 27 June, 21 members of the group that develops Bioconductor announced that they would not develop software for a private file format.

“We wouldn't be willing to do anything that was marginally legal,” says group member Robert Gentleman, a statistician at the Harvard School of Public Health. Researchers working on software in their spare time lack the resources to deal with proprietary file formats, he adds.

Besides the inconvenience to researchers forced to use new software, losing the efforts of a large group working out how best to analyse microarray data would have damaged the field as a whole, Gentleman argues. “The technology isn't mature yet. We need a lot of clever people working on these problems,” he says.

Last week, after conferring with the Bioconductor team, Affymetrix announced that its new compressed format will be made public. “We don't have any trouble with this,” says Kulp. “Clearly, some developers prefer to write their own file readers, which is more work on their part, but has advantages to open-source software distribution.”

“I think that science won,” says Gentleman. But he doubts that the issue will go away: “The clash between open source and commercialization will probably be a constant minefield for both sides.”