By Andy Oram

Over four years of Congresses, Sage Bionetworks has drawn together leading thinkers and doers throughout the fields of genetic research and drug development. For two days each year, the conference floor is colonized by clumps of eagerly networking PhDs from academic, pharma, government, non-profits, biotech firms, and patient advocacy groups–people who often glide from one domain to another within this tight-knit cohort.

A cohort, certainly, we can characterize this group of attendees, sharing as they do a mysterious language drawn from years of research most of us will never understand. But is it a community? That will be tested over the following year as Sage Bionetworks lets go of the Congress. Founder Stephen Friend says it is up to others to create the next Congress, and its success or failure will be a measurement of the sweat and passion that Friend and Sage have put into attempts to build a community.

Why should a reader look further at this struggle among a tiny elite, rather than clicking on the next article? Well, first, if you’re one of the 48% of Americans who took a prescription drug this month, you should be concerned about where new breakthrough drugs will emerge. If you visit this web site because you want a more responsive health care system that can match patients to treatments more quickly and cheaply, recognize that new methods are important nowhere as much as at the foundation of the system where new treatments are discovered. And if you are just curious about the potential for global cross-institutional teams and loose networks connecting experts with ordinary members of the public to find creative solutions to old problems, this article will provide insights.

Don’t get too close, you don’t know what I have

The premise on which Friend founded Sage is that research and drug development have stagnated and cannot progress without more collaboration and data sharing. Therefore, with all due regard for the presentations at the recent Sage Congress on cancer research projects and other individual experiments, the real theme of the conference is in the keynotes about open source, the use of social media, and crowdsourcing. The challenge of this community–if we find that it has indeed become a community–is to analyze and deal with the particular challenges that genetic research and drug development inject into trends toward open collaboration.

It’s hard to determine who has a higher aversion to sharing data: pharma or academia. The starter in the pharma compound is their fear that someone else will discover and patent the billion-dollar drug they’ve been sleuthing after, into which they mix in the worry that publicizing negative trials will lead to their drug’s rejection.

Academics mirror these disincentives, panicking that they’ll lose the race to discovery or that their methods of massaging their data will wither under the harsh light of review. Furthermore, academics are acutely aware of the accomplishments that garner promotion, tenure, and the funding of grants. The mundane tasks of preparing and cleaning data earn them nothing in those areas.

What can detoxify these tendencies to remain isolated? Moral persuasion has limited effect, so the main vector of attack is a growing understanding that tried-and-true research methods are petering out and need some radical rethinking. As I reported from last year’s Congress, the costs of drug development are rapidly increasing, and fewer truly novel drugs are being submitted to regulators for approval over the years.

Stung by evidence that many drugs approved for use are not as safe or efficacious as promised, the FDA has tightened its requirements for Phase 3 testing, which taxes the resources drug companies put into development and further lengthens the time between the patenting of a compound and its actual sale. The “patent cliff” that the drug companies are facing, as lucrative compounds enter the public domain and fewer new ones replace them as sources of revenue, has been widely reported.

Researchers in both pharma and academia have started to cast longing glances at the repositories of data and computer code maintained by other researchers. Just as public health can be judged much better from a nationwide database than from an individual hospital’s case load, genetic research can benefit from larger collections of samples. Still, data is finicky. We’ll look further at some of the complications of data sharing and collaboration.

Pharma companies will continue to hold their cards close to the chest when testing compounds (drug products), but will probably share more data about activities that go on outside their core research, such as tests that validate the efficacy and safety of drugs. Public pressure may convince them to release the data from negative trials, which may save researchers a lot of time when looking for other compounds.

Changing the equations for rewarding research

What incentives can the field offer academic researchers whose careers revolve around publishing papers? One tactic can be borrowed from NIH, which now requires all grant recipients to put their data in public repositories. One manager at the National Cancer Institute lamented that although government agencies require grantees to submit a plan for data sharing, the agencies don’t actually follow up to see that the plan was carried out. So journals can do even better than the federal government in this area.

Further incentives for sharing research data include citations for data sets, which are supported in science by the notion of a Digital Object Identifier (DOI). The field can now track the reuse of data just as it tracks citations in journals.

WikiPathways allows researchers to share information they have about relationships between genes and diseases. As a wiki, the site lets the public edit information about these pathways. But visitors can also search for overlaps and other key relationships.

A platform like Synapse can let writers post drafts of papers that are associated with data stores. Publishers can follow these along and benefit from the free peer review provided by comments. Not only do they make papers better, they help the publisher decide what’s worth publishing. Could these ongoing comments–sometimes called pipelined peer review–substitute for formal peer review? One publisher assured us they would not, but the possibility remains that someday they will, cutting publishers totally out of the distribution of research results.

Room for the patient

Arcane and complex as genetic research is, Sage and its collaborators are committed to bringing patients into the planning process. A number of separate efforts aim at maximizing patient control.

The most familiar kind of patient involvement is the militant advocacy seen in the movement of AIDS victims, chronicled in a talk by HIV researcher Joep Lange. Luckily, both the AIDS professionals and other medical researcher have developed more collaborative relationships with patients as well.

Why should patients determine research goals? Surprisingly, perhaps, genetics researchers in their zeal to uncover causes and treatments often miss what the patients feel is most important to them. Take Fanconi Anemia, a terrible condition involving the blood that gives its victims an average life expectancy of 30 years. One of Fanconi Anemia’s most feared impacts is a higher risk of developing cancer that can drastically shorten the patient’s life. So patients would like researchers to focus on ways to predict cancer and catch it early, not a direction that the research community would discover on its own.

Fanconi Anemia is one of the first areas on which Sage is focusing in its Bridge project. Just in its preliminary stages, Bridge aims to tie patients, doctors, and researchers together. Patients will be abloe donate genetic data and information on their conditions, propose and fund trials, and suggest directions for research. If drafts of papers are posted online during development, Bridge members will be abloe comment on them. The hope is that all this will lead to better research and faster cures.

Bridge will provide a platform for sharing and patient participation that is more open than services such as 23andMe and PatientsLikeMe. In his opening remarks, John Wilbanks laid out three models for medical research. The first is a collection of dynamic but independent companies withholding data for competitive reasons. The second model encompasses relatively recent innovations like 23andMe and PatientsLikeMe. The third is the commons that Sage would like to build. This commons is based on a recognition that progress is not just a series of discrete steps that can be monetized or turned into grant opportunities, but an ongoing exploration involving loosely organized groups of people from many backgrounds.

Consent forms are another area of research under Sage’s microscope. Current forms leave much to be desired. Still in recovery from an era many decades ago when patients were mistreated and kept from a full understanding of what researchers were doing, current consent processes are very heavy-weight and hard for patients to understand. Furthermore, they throw up walls against the kind of data-sharing that Sage promotes.

Meanwhile, popular social networks for patients rely on advertising and other services that aggregate patient data for revenue. These raise privacy concerns for some patients, and make them wonder about the motives of the sites’ owners.

Sage has developed a portable legal consent process that steps the patient through the benefits and risks of sharing (including the risk of re-identification) and lets the patient choose the degree of sharing permitted. Many patients are scared off by the process, quite properly, but those who make it through and decide to sign the consent open up their data to a great many more researchers. Portable legal consent was discussed in depth at last year’s Congress.

When ordinary people get involved in experiments, they can measure personal information (phenotypic data) in their homes and vastly increase the quantities of data available to researchers. Quantified Self enters the lab. But such an innovation, of course, raises questions of how trust data submitted by the patient.

The commitment of Sage to patient control was evident in the range of patient advocates and others who spoke of its importance at the conference. Greg Biggers of Genomera was just one of the experts testifying to the possibilities and potential of bringing patients into the process. The paper describing the results of Sage’s successful breast cancer prognosis challenge offered to develop means for patients to “work alongside Challenge participants” in the future.

The power of positive thinking

It’s easy to provide a forum for public comment and input, but usually unproductive to do so. Whether on the White House’s We the People petition site or Wikipedia, the trick is to find a strategy that encourage participants to contribute positively, not just to carp and tear each other apart.

Let’s review the article and see some of the ways to make participation bear fruit in medical research:

• Helping people form communities with shared goals, such as Bridge.
• Giving patients more of a say in what research is conducted and how it is conducted.
• Safeguarding patient privacy. This is particularly important for populations whose conditions are stigmatizing, such as veterans suffering from neurological damage.
• Providing clean, reusable data and code to researchers, which involved incentives for them to prepare the data and code and describe its provenance. This involves the adoption of standards.
• Requiring data and code to be shared as a condition for funding or publication.
• Running challenges such as the Sage/DREAM breast cancer challenge

What will the coming year see?

The Sage vision for sharing and collaboration is completely in harmony with famous projects outside medicine, including open source software, Wikipedia, Mechanical Turk, and open government initiatives. The means are often also familiar: code repositories in the style of GitHub, challenges, etc.

But many participants at Sage Congress each year need reassurance that these strategies can work. This seemed to be the goal behind lining up such keynoters as Wadah Khanfar, cofounder of Al Jazeera, who talked about openings and transformations in the middle each, Bob Young, founder of Red Hat, and Jennifer Pahlka, founder of Code for America.

Audience members raised typical worries about being open while raising money to get a project done. John Wilbanks, who came from the Creative Commons foundation to lead Sage’s work on portable legal consent, pointed out that making more code and data into free public resources will drastically reduce costs. Still, the attendees are right to be concerned, because research in genetics runs at several million dollars a pop.

Fluctuating buy-in and resistance raises the question whether Sage’s strategy will break through into the mainstream. The conference ended with announcement that surprised even one of Sage’s advisory board: Sage will no longer organize the Congress it has held for the past four years.

Friend’s stated reason for giving up control over the Congress was that Sage is now a player with its own agenda and that the Congress needs to reflect the needs and input of more stakeholders. I suspect the announcement was also meant to put pressure on organizations that have failed to move as fast as Sage. Sage’s work on Synapse, portable legal consent, challenges, and other areas have outstripped the abilities of potential users. It’s time for them to make similar efforts.

But I think Friend has measured his own prospects carefully, and that the announcement was preceded by behind-the-scenes negotiations with key funders and players in genetics and pharma. I expect to see a Congress next year, and to see progress in community-building for which Sage has laid a foundation.

Parts of Sage Congress were videotaped and posted online.

You can also find further coverage from Andy on Sage Congress in the following two articles:
Data Sharing Drives Diagnoses and Cures, If We Can Get There: Part 1
Data Sharing Drives Diagnoses and Cures, If We Can Get There: Part 2

Andy Oram is an editor at O’Reilly Media, a highly respected book publisher and technology information provider. His work for O’Reilly includes the influential 2001 title Peer-to-Peer, the 2005 ground-breaking book Running Linux, and the 2007 best-seller Beautiful Code.