Buetow: A force in the battle against cancer

Biomedical scientists are competitive, highly specialized and — well, let's face it — not given to working easily with researchers in other fields. Cancer researchers are no exception.

But Kenneth Buetow realized that if he could help cancer scientists and clinicians pool their research efforts, perhaps he could accelerate finding a cure for the disease. His idea resulted in the creation of the cancer Biomedical Informatics Grid (caBIG), which researchers now recognize as a significant tool in their fight against cancer.

Buetow, director of the National Cancer Institute's Center for Bioinformatics, conceived of the grid and spent much of 2004 working with cancer centers and companies on building it. Grids are cutting-edge computing networks created to solve large and complex scientific or engineering problems.

"The fact that he managed to get 500 people in the same room and excited and ready to work on this is an important thing — it's very hard to do," said Frank Manion, chief technology officer and senior director for information science and technology at Fox Chase Cancer Center.

Typically, funding for information technology infrastructure to support science is hard to obtain. "Infrastructure often comes from grants that are focused on particular science problems," Manion said. As a consequence, such infrastructure tends to be fragmented.

Like other biomedical groups, the community of cancer researchers and clinicians is unaccustomed to sharing their specialized data and software tools with others in the community. Linking and synthesizing complex datasets from different domains of knowledge is so difficult that it practically requires grid technologies, Buetow said.

Such technologies appear promising for fostering a cross-fertilization of ideas among members of the cancer research and clinical community.

Buetow said he hopes to achieve "semantic interoperability" among the grid's software application centers and data storage centers.

Semantic interoperability refers to standards for translating electronic data so that it is useful to researchers who are not part of the discipline in which the information was collected. It is one of the ways that caBIG differs from other grid computing projects, in which computing centers on a high-speed network simply share computational cycles or disk storage.

"Our goal is to take that up an additional level," Buetow said.

The problem that he is trying to solve is similar to that facing many academic disciplines. Fields in biomedicine have developed in academic isolation, he said. "Immunologists do immunology, genomicists do genomics, trialists do trials, which is fine and has been important and critical to the success of biomedicine to date," he said.

"But the next-generation problems require trialists to use genomics, they require experimental animal modelists to be able to understand what is happening in the clinics, and they require people who are developing novel compounds to understand and have access to gene targets," Buetow said.

Developers for caBIG have begun creating software tools to enforce semantic interoperability and manage an overwhelming amount of new data about genes and proteins.

Significantly, they are designing the tools so that researchers in disciplines other than cancer research can use them, said Russ Rieling, assistant vice president and division manager for innovative informatics technologies at Science Applications International Corp. SAIC programmers have contributed a middleware tool to caBIG.

Among the 20 or more large-scale grid tools in development for caBIG is an open-source tool for collecting and reporting adverse events that occur during clinical trials. Another tool is able to inventory, search for and visualize samples from various tissue banks.

One of Buetow's objectives for caBIG is to create open-source, freely available software using components of the Globus Toolkit and Open Grid Services Architecture Data Access and Integration standards. "Our goal is to move forward at the same time as technologies emerge and mature," he said.

Tools or applications that are not open and freely available must at least be caBIG-compatible. That means there's some level of assurance they'll be able to share data with caBIG applications, Rieling said.

Given the evolving state of grid services technologies, caBIG is ambitious but not impossible, Buetow said. "More than not impossible," he added, "it is pragmatically quite doable."

Part of the National Institutes of Health, the National Cancer Institute (NCI) is financing the entire caBIG effort with about $20 million a year for the first three years. Officials will then evaluate the program to see if it is worth continuing, Buetow said.

Most experts agree that the program is breaking new ground. "We don't believe there is anything comparable to this in the rest of NIH," Rieling said. Officials at other NIH institutes and elsewhere are watching to see how well caBIG succeeds, he said. "If it plays well, they can easily [apply] these technologies in their own institutes."

Although the technical difficulties are daunting, many of the challenges for proponents of caBIG are ordinary business challenges rather than technical ones, Buetow said.

"The hardest part has been the execution of business arrangements so that everyone can feel comfortable with the degree of data sharing, infrastructure sharing and application sharing," he said. "We're clearly challenging people's comfort zones."

The success of such an effort depends largely on openness, Buetow said, "openness in what we're attempting to share, openness in allowing anyone who wants to participate to participate."

As one of its side effects, the caBIG project has brought new visibility to bioinformatics, Buetow said. "We're used to being the people who live in the basements of biomedical research facilities because that's where all the computer centers used to be." In those days, he said, "we got summoned to the upper floors and the boardrooms only when there was a problem to solve."

Pleased with the program's accomplishments, NCI officials offer quantitative measures of caBIG's success. So far, 50 NCI-funded cancer centers are participating; more than 600 researchers are involved; eight broad-based committees are working on architecture, vocabulary and other standards; and 22 special-interest subcommittees are developing specific bioinformatics tools.

In the past year, the committees held more than 285 teleconferences and conducted 10 face-to-face meetings. Their members expect to complete about 75 software tools, databases, prototypes, models and white papers this year.

An NCI fact sheet describes the caBIG program as having potential for redefining how cancer research is conducted and how cancer care is provided.

Many at NCI, including Buetow, expect the caBIG network to accelerate advances in all aspects of cancer research, including new discoveries about prevention, early detection and treatment.

Work on caBIG is driven partly by a deadline set by Andrew von Eschenbach, NCI's director. He said he hopes to eliminate suffering and death from cancer by 2015.

It is a startling pronouncement, but one on which Buetow is focused. Certainly with the bioinformatics grid, Buetow said, "we're bringing the modern tools of [IT] front and center in our attack on cancer."