Letter to Science about wikifying genome information

The March 21, 2008 issue of Science included a News piece stimulated by an open letter from a large group of fungal researchers about using wiki models for improving GenBank. We thought we should send a letter to Science in response.

Please help us edit the text and add your signature if you want to go on record with us.

draft

We noted with interest the recent correspondence from Bidartondo et al regarding a proposal to "wikify" GenBank and the accompanying news article by Elizabeth Pennisi , describing the resistance of GenBank management to the idea. Everyone agrees that current databases are growing at an impressive rate, that this situation is not likely to change in the near future and that incorrect or incomplete annotations can have negative ripple effects beyond the genes annotated by a particular GenBank record. We agree with Bidartondo et al. that the time has come for a serious discussion about alternative models of genome annotation.

We understand that changing the GenBank model should not be done lightly. There are now well-established expectations, both good and bad, about GenBank and RefSeq, and wikifying GenBank could be quite disruptive to the user community. Nevertheless, while our anecdotal evidence suggests that David Lipman’s fears that "It would be chaos." are widely held, it is worth pondering whether the expertise of GenBank's single full-time staff mycologist is a superior on average to the collective, but intermittently applied, expertise of the 250 signers of the Bidartondo et al. letter, plus the many other members of the scientific community who, like us, would have signed the letter if we had known about it.

We have been working for some time on wiki-based systems for gene annotation, including EcoliWiki (http://ecoliwiki.net), the community annotation component of the NIGMS funded EcoliHub project (http://ecolihub.org) and GONUTS (http://gowiki.tamu.edu), a wiki based browser and annotation engine for the Gene Ontology. We regard these, and the similar wiki-based efforts of other groups as experimental, as the actual performance of wiki-based community curation has not been established. However, the scalability, error propagation, and cost problems alluded to in Elizabeth Pennisi's article mean that these are important experiments to perform.

To explore the limits of scaling the wiki approach, we are also currently experimenting with literature-based curation done by students under the supervision of a more experienced scientist as a component of undergraduate research education. Properly designed and conducted, these experiments could change genomics undergraduate education by introducing these future scientists to the huge amount of information already available, and by training them to think and to evaluate the annotated record with that of the published information. When applied to gene families and pathways within and between genomes, this exercise also forces students to think about the evolutionary relationships of the players involved. We think that, if properly conducted, this student-driven engine could change the pace of genome annotation or at least generate a rich testing field for new models for gene and genome annotation. Regardless, an easily accessible public repository of community-driven curated information is urgently needed and the Wiki model offers a good start point.

While these experiments are in progress, perhaps GenBank could support an arms length relationship to wikis via the LinkOut service. We look forward to participating in the ongoing discussion of the future of how community curation could be added to GenBank. In the meantime, we encourage the signers of the Bidartondo et al. letter to help us build wiki-based fungal curation of gene function via GO at GONUTS.

---