Spots, bands and peaks. Readers could be forgiven for being hard pressed to make a connection but, to the up-and-coming biologist, resolving spots and bands on polyacrylamide gels and reading peaks from a mass spectrophotometer lie at the heart of high-throughput biology. Technology and robotics lead the protein-biology revolution and, as quantity gains credence, biologists must take control of the avalanche of information being presented.

Proteomics is defined as the identification and quantitative expression analysis of every protein encoded by a genome, characterizing protein–protein interactions and assigning these protein complexes with function. Biologists have undertaken large-scale screens to identify every protein expressed on a single chromosome and determine the complement of proteins that defines a complete biological process. But in the rush, the golden rules of biology that allow for easy comparison, exchange and verification of data are being overlooked.

Hordes of data have been analysed, but it is difficult to compare data sets because they are derived under different conditions, presented in different formats and benchmarked against different reference sets. This information must be accommodated in 'interoperable' repositories, and many databases are being set up for this purpose. This requires distributed databases from the outset that include common machine-readable tags for the species, strain or tissue used, say. By agreeing a minimal set of common standards, the proteomics community will ensure that their many databases can communicate with each other in the same language, and be queried as if they were a single resource.

In April this year, the Human Proteome Organisation founded the Proteomics Standards Initiative (http://psidev.sourceforge.net), which aims to define the standards required. This will be a huge challenge, not least because, as one computer scientist once quipped, “standards are so important that everyone wants their own”. But if the full promise of proteomics is to be realized, the community and the bioinformaticians that service it need to pay more attention to the often thankless task of standard setting. Only then will it be possible to find order in the labyrinth of biomolecular networks.

· Launched this week: the AfCS-Nature Signalling Gateway - see http://www.signaling-gateway.org/