Genome Database Booms As Journals Take The Hard Line

For researchers in such fast-growing fields as gene sequencing and crystallography, the end of traditional scientific publications may be in sight. Quietly, over the last year, a trial program in the electronic submission of data has laid the foundations for a revolution that may someday replace the research journal as the vehicle for scientific communication. In an agreement With GenBank, the main U.S. gene sequence database, nearly a dozen journals are refusing to accept papers with sequence data unless the authors can supply an “accession number,” proving that they have already submitted the information to the database. Other journals are strongly encouraging such prior electronic submissions and are likely to make the rule mandatory in the future. The policy is aimed at a growing problem among electronic databases in hot fields. Swamped by a flood of published data, the databases are quickly falling behind, and losing much of their usefulness for scientists intent on keeping up with the latest research in their field. For the gene sequence journals, the policy is intended to deal with the failure of researchers to submit their data routinely to GenBank, set up at Los Alamos National Laboratory by the Department of Energy (DOE) and the National Institutes of Health as the chief repository for such infor mation. Many researchers saw the submission of their sequencing data to GenBank as just one more item on an already overcrowded list of responsibilities, and voluntary compliance was low. As a result, GenBank was forced to employ a half-dozen clerical workers, who entered data into computers from some two dozen journals. It takes an average of one hour to type in the pages of variations of the letters A, G, C, and T—abbreviations for the four possible chemical building blocks of the DNA molecule—that typically compose a journal article on a new sequence. Each sequence must be entered twice and checked by computer to spot clerical errors. But the real problem wasn’t the additional labor; it was the resulting delay in making the information available to the,community. The operators, facing mounds of genetic information, were as much as 12 months behind in recording the data that were appearing in print. That backlog has shrunk to as little as a week in the year since the trial has been under way. Nearly two thirds of the entries are arriving electronically or in computer-readable form, and GenBank officials hope to raise that number to 80% by the end of the year. At that point, says program director Paul Gilna, the information in GenBank will be almost as timely as the research itself. Warm Reception Surprisingly the new policies have elicited few complaints from researchers. Because most of the data are destined to be published anyway, scientists rarely reveal anything that they would rather keep to themselves. And although the new rules mean that sequence data must be submitted to GenBank months before publication, there appears to be little concern that competitors could benefit from such an advance look. Indeed, although GenBank allows researchers to “embargo” their data until their paper is published, only about a quarter of those who submit material request it. GenBank’s success with direct data submission is being watched closely by scientists and database plannersin other fields as well. Crystallographers, for example, are embroiled in a heated debate over the issue of forced submission of databases because of the sensitive nature of their data. In X-ray crystallography, data typically consist of long lists of three-dimensional coordinates that describe the atomic structure of large molecules, like proteins or nucleic acids. The information is critical to such industrial researchers as drug designers, who want to create compounds that will target the mole- cules, or genetic engineers, who wish to modify them. Coordinates can represent big bucks for crystallographers—but only as long as they are the first to mine their trove. As a result, many have refused to publish essential parts of their data, or have obscured them enough to obtain a head start on analysis. Sequence data, on the other hand, are less closely tied’ to industrial products and profits, and geneticists seem to share few of the complaints raised by crystallographers. Questions Remain Nevertheless, several issues remain unresolved A growing number of submissions (about 10% to 15% of the incoming sequences) may never be published in any other form. “Some scientists submit data into the database just to get a ‘time stamp,’ as a kind of copyright issue,” Gilna says. Others send in their data as a service to the community, hoping that others will do the same. In fact, a new menu-driven software package called Authorin is being distributed freely among scientists to simplify the process of electronically submitting gene sequences. The problem with such data is that they have not been peer reviewed. Because GenBank traditionally has conducted only minimal checks for accuracy, there is some concern that the new policy may lead to errors in the database. “If they say it codes for a certain protein, we check that, but we can’t check everything,” says Gilna. Adds David Lipman, director of NIH ‘s National Center for Biological Information (NCBI), “Do you really believe that the person who submitted the unpublished data was as careful in their analysis as if they had submitted it to a journal? I don’t.” In part to address this. concern, GenBank has asked 15 prominent researchers from around the country to become database "curators." Equipped with the necessary computer hardware and software, the curators will regularly review new database entries, adding comments, grouping related research, and watching for suspect data. “It’s both a value-added program and a quality-control program,” says Gilna. He admits, however, that bad data will inevitably slip through. To some scientists, the issue data integrity—and the need for accuracy—looms larger as the database comes closer to becoming the electronic embodiment of a physical reality. Says Robert Robbins, director of the National Science Foundation biological database program: “There is a subtle but very real difference between a database that purports to be about scientific literature, and one that purports to be about the real world—the human genome, for example. (The latter] is obliged to perform a level of quality control beyond a literature database.” More To Come However, nobody expects GenBank’s current incarnation to be the last word on the subject. NIH’s NCBI is preparing a database that will be based largely on unsolicited and unpublished data (The Scientist, Sept. 4, 1989, page 3). Another NCBI project is meant to allow the database to evolve in line with the science itself. Using unusually flexible software techniques, it is intended to adapt as scientists’ understanding of gene sequencing matures. “Current databases are roller skates,” says Mark Pearson, director of Du Pont’s molecular biology program and a member of the NIH/DOE database advisory committee. “What GenBank and NCBI are moving up to is a bicycle. But we’re still looking for automobiles.” Whatever the mode of travel, researchers say GemBank’s experi- ment with enforced submission is transporting databases into a new realm. As the links between scientists and science databases get closer, researchers are starting to reexamine the role that information storehouses should play in the scientific enterprise. “Will databases be just another level of journal, or are we on the eve of a new form of scientific literature?” wonders NSF’s Robbins. The lessons learned from the new generation of genome databases may provide the answer.

Genome Database Booms As Journals Take The Hard Line

Interested in reading more?

Become a Member of