In June 2000, US President Bill Clinton and British Prime Minister Tony Blair unveiled what amounted to a "rough draft" of the deciphered complex makeup of human genetics, an essential milestone to cracking the genetic code that makes up human life.

Work on the mapping of the human genome, whose completion was announced in April 2003, was heavily dependent on advanced computing for the data intensive task of mapping the sequence of 3 billion base gene pairs in humans.

Ironically, getting that genetic data into the hands of biomedical researchers has created another major computer quandary: the need for even more advanced systems that can keep up with an increasing number of disease subcategories being discovered through genetic research.

The US National Cancer Institute took on the task of addressing that issue in 2003 by launching what it calls the largest IT project in the history of biomedical research. The NCI created what is, in essence, a world wide web of cancer research.

The new Cancer Biomedical Informatics Grid, or caBIG, promises to help researchers, physicians and patients to better share more-detailed information about diseases and thus speed the development of new drugs and treatments for them.

The government-funded effort cost about US$20 million a year, the NCI said.

To date, 42 of the institute's 63 national cancer centres are either linked to the caBIG grid or are installing the necessary infrastructure to participate. Many are already building applications that can be shared by members of the grid.

The need for wider data sharing became obvious as genetics research found more sub-categories of cancers that would require specific treatment methods.

Cancer researchers quickly saw the need to assemble as much information as possible to help in the development of new disease-specific treatment options. So, to broaden the number of data sources, the NCI has begun expanding the grid to include the community hospitals and physicians that treat 80 per cent of US cancer patients.

Interoperability

Project backers said that researchers decided early on to focus on improving interoperability rather than forcing research organisations to standardise on expensive new IT systems and software.

To accomplish that, the developers used the Globus Toolkit, a set of open source tools for building grid systems and applications that run on top of web services that are open for anyone with a node on the system. The Globus tools are distributed by the Globus Alliance.

Developers also created a collection of tools that serve up semantic descriptions of vocabulary and data so that both humans and machines can interpret data from dissimilar systems. And a common security model was built to allow research centres to run caBIG as a distributed infrastructure that lets each participant create individual policies to determine who can author or access data.

In addition, Ken Beutow, director of NCI's Center for Bioinformatics, said the NCI has set up "workspaces" -- groups of people that meet regularly to discuss specific domains of work like tissue banks and pathology tools. The workspace groups provided input on building the common vocabularies and data elements, he said.

Robert Annechiarico, director of cancer centre information systems at Duke University, which has already helped build applications for the grid, said that creating the common data elements is especially important in the academic world. "Academic medical centers are a community of fiefdoms bound together by a common parking problem," he said.

Researchers at Duke contributed to the development of two caBIG applications, the Cancer Central Clinical Database and the Cancer Central Clinical Participant Registry.

The latter application, a web-based tool for managing clinical trials data across multiple cancer centres, can provide researchers with access to records about patients suffering from one of the new subcategories of cancer.

"Where I might see five patients a year with a particular disease, now I can see 50," Annechiarico said.

Duke is using the former application in a $6.8 million research project, funded by the US Department of Defense's Breast Cancer Research Program, to study how genomic profiling can be used to guide treatment plans for women with newly diagnosed breast cancer, he said.

In addition to expanding access to specific data sets, caBIG can increase the safety of clinical trials for patients, according to Warren Kibbe, director of bioinformatics at the Robert H. Lurie Comprehensive Cancer Center at Northwestern University in Illinois.

For example, he said, development of a caBIG clinical trials management application would allow researchers to determine the adverse effects of a single medication used in multiple clinical trials. "That is one example of how caBIG is starting to touch patients in a way that hasn't been possible," Kibbe said.

The open-source Patient Study Calendar application now in development at the centre will be used for patients on clinical trials. Among other things, the application will be able to tell patients how much medication to take and when.

The single application could define patient management parameters, eliminating the problem caused when people with different types of training - a surgeon versus an oncologist, for example - may interpret rules differently, Kibbe said.

Implementing caBIG has not been without challenges, according to an NCI-commissioned review of the project that was released late last year.

The report found that over the life of the effort -- from 2003-2007 -- developers have not focused enough on the needs of end users, and have too often released buggy products.

Beutow said the report prompted the NCI to "redouble" its efforts to provide better technical support to users. The agency now sends updates on the program to user e-mail lists, has created web sites with caBIG information, and launched a telephone help line to provide technical support to users.