www.nature.com/nature Vol 455 | Issue no.

7209 | 4 September 2008

Community cleverness required

Researchers need to adapt their institutions and practices in response to torrents of new data and need
to complement smart science with smart searching.

he Internet search firm Google was incorporated just 10 years credit, then, to those in the vanguard of interoperability. In biology,

T ago this week. Going from a collection of donated servers for example, the Gene Ontology Consortium has spent the past
housed under a desk to a global network of dedicated data decade devising consistent descriptions of gene products in differ-
centres processing information by the petabyte, Googles growth ent databases. Meanwhile, the Mouse Genome Informatics resource
mirrors that of the production and exploration of data in research. is a good demonstrator of complexitys challenges and solutions.
All of which makes this an apt moment for this special issue Funding agencies have been slow to support data infrastructure and
of Nature, which examines what big data sets mean for contempo- this is one cultural shift that needs to accelerate although recent
rary science. efforts by the US National Science Foundation and Germanys DFG
Big, of course, is a moving target. The portability of the tens of are a good beginning. But above all, such standards require support
gigabytes we carry around on USB sticks would have seemed like from researchers, who should adopt them and deploy them consist-
fantasy a few years ago. But beyond a certain point, as an increasing ently. This takes a degree of intellectual and practical commitment
number of research disciplines are discovering, the vast amounts of to what can seem like tedious bookkeeping.
data are presenting fresh challenges that Researchers need to be obliged to
urgently need to be addressed. EDITORIAL document and manage their data with
The issue is partly a matter of the sheer 1 Community cleverness as much professionalism as they devote
scale of todays data sets. Managing this required to their experiments. And they should
torrent of bits has forced more and more receive greater support in this endeav-
fields to move to industrial-scale data NEWS our than they are afforded at present.
centres and cutting-edge networking 8 SPECIAL REPORT The next Google Those publicly funded databases that
technology (see page 16). But the data Duncan Graham-Rowe have taken on preservation responsibil-
sets are also becoming increasingly PARTY OF ONE ities, such as GenBank and UniProt, are
complex. As researchers study the inner 15 Data wrangling only a small part of the data landscape.
workings of the cell, for example, they David Goldston Universities and funding agencies need
now gather data on genomic sequences, to provide and support curation facili-
protein sequences, protein structure and NEWS FEATURES ties, tools and training.
function, bimolecular interactions, sig- 16 Welcome to the petacentre As is amply highlighted in this issue,
nalling and metabolic pathways, regu- Cory Doctorow all of these worthy aims require incen-
latory motifs on and on. No wonder 22 Wikiomics tives. These include pressure from, and
even the smartest scientists turn with Mitch Waldrop recognition through, journals. Nature
relief to advanced data-mining tools, and its sister publications have always
online community collaborations (see COMMENTARY worked closely with those develop-
page 22) and sophisticated visualization 28 How do your data grow? ing databases and standards, and we
techniques (see page 30). Clifford Lynch remain committed to continuing such
Sudden influxes of data have trans- community collaborations. Incentives
formed researchers understanding of
BOOKS & ARTS also include recognition of impactful
nature before even back in the days 30 Distilling meaning from data informatics by peer committees and
when computer was still a job descrip- Felice Frankel & Rosalind Reid research-rating exercises.
tion (see page 36). Unfortunately, the ESSAY Above all, data on todays scales
institutions and culture of science remain 36 The Harvard computers require scientific and computational
rooted in that pre-electronic era. Taking Sue Nelson intelligence. Google may now have its
full advantage of electronic data will critics, but no one can deny its impact,
require a great deal of additional infra- FEATURE which ultimately stems from the clever-
structure, both technical and cultural 47 The future of biocuration ness of its informatics. The future of sci-
(see pages 8, 28 and 47). Doug Howe, Seung Yon Rhee et al. ence depends in part on such cleverness
The lack of standards, for instance, again being applied to data for their own
For podcast and more online
confounds many a researcher seeking sake, complementing scientific hypoth-
extras see www.nature.com/news/
to harness the diversity of knowledge specials/bigdata/ eses as a basis for exploring todays infor-
now available on any chosen topic. All mation cornucopia.