Imagine that for selected cancer patients, biopsies are taken before, during and after treatment, made anonymous and the analyses stored promptly in an accessible fashion. These biopsy samples are subjected to gene-expression and proteomic analysis, and these molecular data are also stored accessibly. Imagine also that the patient's data can readily be compared with those from other trials. And imagine that one can drill down into clinical and other databases in an intelligent search in hours rather than months. One end-point might be the rapid identification of individualized molecular profiles correlated with sensitivity or resistance to therapy.

This vision requires common standards of data storage at each level of investigation, new frameworks for cross-referencing terms and their biological contexts (‘ontologies’) between disparate types of data, and new bioinformatics tools to make it all practicable. The benefits? Quicker routes to identifying patients' individual characteristics that make one treatment more appropriate than another; easier integration of genomics research into clinical trials; and much readier access by basic molecular and cell biologists to the early lessons that can be drawn from even a few patients, as well as from large-scale, randomized clinical trials.

Much of this is beginning to be realized. The US National Cancer Institute last week launched its Cancer Biomedical Informatics Grid project, bringing together the institute's Center for Bioinformatics (NCICB) with translational centres and clinical-trial cooperative cancer centres in the United States (see http://cabig.nci.nih.gov). The programme, which costs $20 million a year, will yield networks of clinical-trial information, tissue data, ontological development and integrative tools that give researchers ready access to data.

In close collaboration, Britain is embarking on the same route. The strategic body that represents most of the relevant UK funding agencies, the National Cancer Research Institute (NCRI), is this week announcing in Nature (see http://www.cancerinformatics.org.uk) a framework for similar developmentsFootnote 1.

Although the NCRI efforts will involve centres of excellence, they currently lack the dedicated budget of the NCICB. It is therefore critical that funding agencies collectively set aside millions each year for these developments, and that a combination of strategic leadership and peer review is established, as in the US programme.

One unique part of the UK landscape is both a boon and a potential weakness: the National Health Service (NHS). Its existence gives Britain a greater ability to focus its efforts on standards and storage than the patchwork of state and privately funded health services in the United States allows. But information technology in the NHS is in dire need of development. And whatever happens, it is essential that suppliers of the technology do not gain control over access to public data, as has happened in other contexts.

More positively, despite the need of researchers to protect not only their patients but also their competitive interests, the leaders of these bioinformatics initiatives have been gratified by the positive attitudes to data sharing encountered so far on both sides of the Atlantic. Next, and sooner rather than later, comes the challenge of extending cancer bioinformatics collaboration across the disparate research and health systems of Europe.