The Disease Machine

During a six-year stint as head of cancer research at Merck, genetics entrepreneur Stephen Friend tried to inject a new kind of biology, synthesized from supercomputers and reams of biological data, into a staid drug maker as it struggled to follow up breakthroughs in treating heart disease, osteoporosis and infection.

Then, in February, Merck ( MRK – news – people ) announced Friend was leaving to conduct an “ambitious and promising” new project. Over the past few months, in drips and drabs, Friend has laid out his plan: to use software tools developed at Merck to pool data from drug companies, universities and patient groups into one of the most comprehensive computer models of human biology ever created. The idea is that disease isn’t just a collection of symptoms, or even a single mishap. It’s a rippling disruption of the body’s network. Figuring out what went wrong is the way forward for biology. The name of the effort is Sage Bionetworks.

“This fundamentally changes how biologists do an experiment,” says Friend. “Disease biology will not make sense unless we can do this. But instead of reacting as if it’s sci-fi, people are saying ‘yes, this is obvious.’ We need coherent blocks of data, we need models of disease.”

Sage comes at a time when the boundaries that have confined medical research are falling. One genetic sequencing effort, the Personal Genome Project, is offering all sorts of medical and personal data about its research subjects in the hope of fostering advances. Patient groups like the Cystic Fibrosis Foundation, the Multiple Myeloma Research Foundation and the MPD Foundation fund researchers with the explicit demand that they share data with one another. One of these groups, CHDI (formerly the Cure Huntington’s Disease Initiative) has provided start-up funding to Sage. “There’s a lot of data that’s floating around already,” says CHDI President Robi Blumenstein. “We want to have a seat at the table to get a better understanding of how we can apply this to Huntington’s.”

Even drug companies are starting to share more and more data. In 2007, Novartis ( NVS – news – people ) made information about genetic links to diabetes publicly available. Eli Lilly ( LLY – news – people ) has made sharing research a fundamental strategy, and says it might be interested in something like Sage. Merck itself is currently Friend’s biggest backer. The drug company gave Sage $5 million in start-up funding and donated the fastest supercomputer in the drug industry to his efforts. About a dozen researchers who worked at a Seattle lab Merck closed have now moved over to Sage.

Pharma is opening up because its record of inventing things is abysmal. Last year, 22 new drugs hit the market, half as many as a decade ago. The big test for Sage will be whether creating what Friend calls “a pre-competitive space”–a sandbox in which everybody can do research–will actually end the drug discovery drought.

Read All Comments

Friend started his career as a doctor, treating children with cancer. But then he started thinking about developing drugs. In 1996, he co-founded Seattle-based Rosetta InPharmatics, which aimed to use computers to filter genetic data. Rosetta had a $100 initial public offering in 2000 and was bought by Merck for $620 million in 2001. Rosetta became Merck’s genetic research arm, and Friend was moved east to the company’s West Point, Pa., laboratories. In 2003, he took over basic cancer research, where Merck had lagged and where the genetic approach was likely to hold the most promise.

Enter a Rosetta researcher named Eric Schadt, who had joined Rosetta in 1999. Schadt grew up in a religious fundamentalist family thinking he would be either a farmer or a mechanic. He joined the Air Force’s physically demanding rescue team, and badly dislocated his shoulder while repelling off a cliff. He’d scored very high on his aptitude tests, and during his recovery was exposed for the first time to higher education. He became a mathematician and molecular biologist.

Schadt started working on software that would pool different types of biological data–clinical results, drug responses and genetic data–into a single model of a biological network. In 2003, he published data showing his system worked in corn. Science hailed a 2005 Schadt paper as evidence that looking at the body as one vast network might yield fruit. In 2008, Schadt and other Merck researchers used the approach to identify new genes for obesity and diabetes. (See “All Connected.”)

Friend was using Schadt’s software to work on cancer drugs. (See “Merck’s Free Radical.”) But he and Schadt were starting to realize that there was no way that Merck alone would be able to aggregate enough data to make the software really work. At the same time, Merck was cutting jobs as part of an industry-wide slump, and the Rosetta facility was going to be shut down. The technology they had brought in had helped build Merck a pipeline of early, experimental cancer drugs, and informed its research. But Friend says he regrets that he didn’t get his new approaches more deeply ingrained into the research effort there.

After starting Sage, Schadt took a second job as chief scientific officer of Pacific Biosciences, the gene sequencing start-up. He is splitting his time between the two efforts, which he sees as part of the same thing. Stephen Turner, the sequencing company’s founder, says that both of Schadt’s jobs amount to trying to understand “the grammar of genetics.” Meanwhile, Friend has been getting Sage, which he sees as an international effort, going, first from a houseboat, and now from a new house. He’s back in Seattle, in offices at the Fred Hutchinson Cancer Research Center. The idea is to get data in, than open up the system to the world over a period of a couple years.

“People would build what they want to build on the platform in a way that has not been possible to do in biology,” says Friend. Adds Schadt: “No one group or person understands what’s happening in Wikipedia. I see biology undergoing that same transformation. A big change is coming.”