Metagenomics sprang from advances in sequencing technology, and continued improvements are providing data in quantities unimaginable a few years ago. But without concerted efforts, the amount of data will quickly outpace the ability of scientists to analyse it. The September Editorial of Nature Methods (6, 623; 2009), ‘Metagenomics versus Moore’s law’ draws attention to some articles in the same issue of the journal that illustrate some of the dangers and problems, as well as the solutions that are being sought.

Three years ago, the Editorial continues, the first two second-generation metagenomes were reported at less than 40 megabases each. Now, there are more than 4,000 sequenced metagenomes that would take years or tens of years to analyse (depending on the processing power used). Major initiatives are needed to avoid metagenome-analysis gridlock: according to the Editorial, funding agencies need to increase support for data analysis; and the community needs to improve data-sharing through standards and centralized coordination and by aggregating computationally intensive operations. The conclusion:

“This summer, after discussions at the International Conference on Systems for Intelligent Molecular Biology, community members formed the M5 (metagenomics, metadata, metaanalysis, multiscale-models and metainfrastructure) Consortium under the roof of the Genomics Standards Consortium to devise a solution to the coming gridlock. Their proposed ‘M5 Platform’—to be announced later this year—deserves the support of the community, funding agencies and those who hold the keys to the high-performance computing centers. Unless major efforts are taken immediately, researchers will find they have a wealth of data but no way to interpret it.”

Readers’ comments and discussion of this Editorial are welcomed at Methagora, the Nature Methods blog.

This is the text of a Correspondence by Richard Behringer of the University of Texas, published in the current issue of Nature (460, 324; 16 July 2009).

I was disappointed by the view expressed in your Editorial ‘The sharing principle’ (Nature 459, 752; 2009 – free to read online) that the mouse community does not share its strains. This is untrue. Most labs are very collegial, spending a considerable amount of time and effort on distributing their mouse strains. Although there are a few labs that withhold distribution, any community may contain such individuals. The fact that a mouse strain is not found in a repository does not mean that it is not being shared.

I was also puzzled by the conclusion of May’s CASIMIR workshop, noted in the Editorial, that “the sharing problem urgently needs resolution” with regard to international mouse gene-knockout projects. Such mutant alleles will mostly be archived as embryonic stem-cell lines. Readers should also realize that repositories cannot keep all their mouse strains live ‘on the shelf’: most strains are frozen. The cost for a user to have a strain thawed is thousands of dollars and it takes many months before the recovered mice become available. This is a big disadvantage for labs on tight budgets. With regard to funding agencies: in grant proposals to the US National Institutes of Health, for example, applicants are required to write a ‘resource-sharing plan’ that includes genetically modified mice.

It was suggested that sharing avoids duplication of effort. But it is essential that more than one group generates mutations in the same gene as a crosscheck. No two labs generate the same allele, and every geneticist knows that the expression of different alleles can lead to very distinct phenotypes.

Your claim that sharing mice “has never been easier” is questionable, considering all the paperwork, health certificates, veterinary screens, special serology screens, costs, time and logistics involved. This is quite different from uploading DNA sequences in the comfort of your office.

It would be great if funding agencies supplemented grants involving the generation of mouse strains to cover the costs of sending the strains to a repository. In these tough financial times, that seems unlikely.

Department of Genetics, University of Texas, M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas 77030, USA.

Nature journals’ policy on data and materials availability.

A universal tagging system that links data sets with the author(s) that generated them is essential to promote data sharing within the proteomics and other research communities. The July Editorial in Nature Biotechology (27, 579; 2009) reports the results of the journal’s survey of author compliance in depositing proteomics and molecular-interaction data underlying the papers they published. The editors found that even authors who are proponents of data deposition are not making data available in all of the papers they publish. Inhibitory factors include data quality and the user-unfriendliness of some databases. The Editorial concludes:

“One option would be to provide researchers who release data to public repositories with a means of accreditation. This would take the form of a universally standardized tag for data that could be searched and recognized by both funding agencies and employers. An ability to search the literature for all online papers that used a particular data set would enable appropriate attribution for those who share. In essence, the tag would be a digital object identifier (DOI), currently best known for its use in unambiguously identifying papers online.

Similar to citation information about publications, citation information about a researcher’s data DOIs could be gathered by funders assessing future support and used by institutions in performance evaluation. Researchers who disclose data sets that subsequently prove particularly useful to the community would end up with highly cited data DOIs, and could thereby be rewarded accordingly.

Such a system would not solve all the problems slowing data disclosure in proteomics and elsewhere. But it would provide greater incentive than the present system of evaluation, which is skewed almost exclusively to publications in high-profile journals and citation metrics. Data DOIs would not only enhance a researcher’s reputation but also establish priority of data generation. Most important of all, they would provide a way to acknowledge the time and effort individuals must invest in sharing data, which ultimately benefits the scientific community as a whole.”

See also a Correspondence in the same issue of Nature Biotechnology (27, 597-598; 2009): PRIDE Converter: making proteomics data-sharing easy, by Harald Barsnes, Juan Antonio Vizcaíno, Ingvar Eidhammer and Lennart Martens, a collaboration between the University of Bergen and the European Bioinformatics Institute.

Nature journal policies on data and materials availability.

This is the text of one of the Editorials in the current issue of Nature, The sharing principle (Nature 459, 752; 2009):

Back in 1996, human-genome scientists signed up to the Bermuda agreement to share their data without delay. Since then, the sharing principle has entered the mainstream — it now applies to all genomic data generated using public funding, as well as to all the relevant resources cited in publications.

But this principle is not universally observed for genetically modified mice, designed as vital resources in the quest to unpick basic biological mechanisms or to model human disease. The size of the problem is unclear, but existing surveys, combined with extensive anecdotal evidence, suggest it is substantial. In April 2006, for example, scientists at the US National Institutes of Health found that nearly 4,000 unique mice strains had been created, yet barely 700 had been placed in a repository.

Some scientists say they do not have the time nor money to breed and distribute their mice, or even to send the animals to publicly funded mouse repositories such as the European Mouse Mutant Archive, the Jackson Laboratory in Bar Harbor, Maine, and RIKEN BioResource Centre in Ibaraki, which would do those chores for them. Others claim that the careers of young and vulnerable researchers (or old and vulnerable researchers) could be harmed if they lost their exclusive access to a resource they made for their own research projects. Or, they say their institution’s technology-transfer offices or companies will not let them part with mouse strains that could perhaps be made to turn a profit.

Such attitudes were noted with concern last month at a workshop in Rome hosted by CASIMIR, a European Union project to coordinate and sustain mouse resources internationally (see background documents). The workshop brought together representatives from funding agencies, publishers and the mouse repositories from Europe, the United States and Australasia. They concluded that the sharing problem urgently needs resolution — not least because international projects to systematically generate mouse lines deficient in each gene in the genome will generate thousands of new strains in the next five years or so.

To solve the problem, however, journals and funding agencies must take a tougher line. The Nature journals are among the very few actually requiring that authors use established public repositories wherever possible as a condition of publication. Most journals simply ‘encourage’ their authors to make mice used in their publications freely available to other laboratories, or ‘suggest’ that the mice be deposited in repositories. Funding agencies similarly prefer such cajoling terms as ‘encourage’ in their policies on sharing mouse resources, and rarely police the outcome.

Journals should now require researchers to place their mice in repositories as a condition of publication. And funding agencies should require repository plans to be included in all grant applications that are likely to generate new mouse strains. Part of the grant money should be reserved for this task and final reports or evaluations of the grants should refer to the repository used. The repositories themselves should help the journals and funding agencies by finding a way to generate a unique accession number for each mouse strain.

The sharing principle allows biology to progress efficiently. It avoids duplication of effort and allows different laboratories to use the same tools. It is essential that scientists sign up to it. Sharing mice has never been easier — the repositories around the world are efficient and professional, and they are coordinated. Just a few changes in the modus operandi of key institutions could ensure that the makers of mice will have no possible excuse not to use them.

Nature journals’ policy on availability of data and materials.