Science magazine has released details of the terms under which it plans to publish Celera Genomics' paper on the human genome — and drawn sharp criticism over the limits that the terms set on data access.

Celera and Science have agreed that publicly funded scientists can download up to one megabase of data without signing a material transfer agreement. But privately funded users must sign such an agreement stating that they will not commercialize their results or redistribute the data.

The arrangement has potentially profound significance, some researchers say, because it could set a precedent for scientists to publish papers without unencumbered access to supporting data.

“What will happen if someone else from the academic sector says, 'I have an interesting result to report, but I can't give you all the data'?” asks Harold Varmus, president of the Memorial Sloan-Kettering Cancer Center and former National Institutes of Health director. Varmus was one of several scientists who advised Science on the terms. But the advisers are not all in agreement with the announced policy, he says.

Bruce Alberts: deal won't work for private sector.

When the agreement was announced, Bruce Alberts, president of the National Academy of Sciences, issued a brief statement of support, saying that it may serve as a way to prompt more companies to make their genomic data available.

But this week, Alberts — who says that the statement reflected his own views, not those of the academy — backed off from this. In an interview, he said that the deal could work for publicly funded scientists, but that the terms for private ones appear unworkable. “The data should be available both to the public sector and private,” he says.

The agreement has drawn vocal criticism from some researchers. Ewan Birney, team leader of genomic annotation with the European Bioinformatics Institute in the UK, co-authored an open letter to bioinformaticians attacking the arrangement. He says it is problematic for computational biologists, because they work with large data sets, whose transfer is restricted by the agreement. “This deal is bad for bioinformatics, but palatable for single gene biology,” Birney says.

→ http://www.sciencemag.org/feature/data/announcement/genomesequenceplan.shl