Open Humans Aims to Be the Social Network for Science Volunteerism
By Aaron Krol
April 9, 2015 | Jason Bobe, Executive Director of the nonprofit PersonalGenomes.org, is a serial research participant. He is a proud donor of skin cells to the New York Stem Cell Foundation, and fondly remembers swabbing his cheek with super glue to provide facial mites for a study at the North Carolina Museum of Natural Sciences. As a member of the Harvard Personal Genome Project (PGP), he has also made himself a subject of one of the most comprehensive personal exams in the world, sharing his genetic data and filling out exhaustive surveys about his health and physical traits.
But Bobe recognizes that, as a frequent volunteer for science, he’s part of a small minority.
“You’d be amazed,” he tells Bio-IT World. “I go to health conferences all the time, and I almost always ask if I’m giving a talk, how many of you in the audience are currently actively enrolled in at least one health research study? And even in an audience enriched with researchers, it’s usually around 5 or 10%.”
At PersonalGenomes.org, Bobe is working to drum up much higher levels of public engagement with science. That mission, he feels, will be best served if scientists realize that engagement is a two-way street. Rather than collecting data and keeping it locked up in their own records, scientists could offer research participants copies of any personal data they generate, whether just to satisfy their own curiosity or to share information with more studies in turn.
“We’re trying to basically establish a new normative behavior here, where researchers start to share more data,” Bobe says. “We think that’s essential for people to be able to make informed sharing decisions, and also going to be a key driver of engagement in health research.” It’s a tactic that many of the studies for which he volunteers have adopted: as part of the Harvard PGP, for instance, he owns a personal copy of all the genetic data he has provided to the project.
This March, PersonalGenomes.org launched Open Humans, a sort of social media site for participants in scientific studies. Backed by $1 million in grants from the John S. and James L. Knight Foundation and the Robert Wood Johnson Foundation, Open Humans aims to create a public meeting ground for citizen scientists. Through their Open Humans profiles, members can gather together datasets from multiple studies they participate in, and share or publicize that data however they choose.
The project includes three research partners at launch: the Harvard PGP; American Gut, which explores the bacteria and other microbes living in the human digestive tract; and GoViral, which collects samples from people with colds and other respiratory illnesses to track the infection patterns of viruses. (Bobe is a participant in all three.) These partners have a history of cooperating and sharing data with one another, through the annual GET Labs event in Cambridge, Mass., where PGP members get together to expand their participation in citizen science — the same gathering where Bobe swabbed himself for face mites last April.
Open Humans aims to do more than unite a few specific studies, though. Ultimately, Bobe and his colleagues hope to give research participants the power to open up formerly closed science, drawing together diverse datasets that are more meaningful together than apart. The project is an acknowledgement of ordinary people’s growing power to contribute to scientific knowledge. It’s also a response, in the age of personalized medicine, to scientists’ need for multiple, complementary sources of data to unravel the complexities of human biology.
The “Democratic” Approach
It’s no accident that the Harvard PGP is the nucleus of the Open Humans project. As part of its mission to draw new connections between human genetics and health, the PGP sequences participants’ whole genomes and collects information on dozens of health traits, data that provides a useful core for an Open Humans profile. More importantly, the PGP was one of the first studies to experiment with data governance strategies like equal access and public sharing, making it a template for some of the more sensitive features of the Open Humans network.
When the PGP was created in 2005, recruitment for the project was hampered by severe restrictions on who could join: at the direction of Harvard’s Institutional Review Board (IRB), every volunteer for the PGP needed a master’s degree in genetics or equivalent training to participate. These volunteers weren’t looking to lend their time in the laboratory; they simply wanted to be research subjects in a study they found exciting. But the PGP founders wanted to experiment with “equitable data sharing,” in which research participants get to own and retain copies of their personal data — including, in this case, their genomes. The university wasn’t sure how to accommodate this vision.
“This was ten years ago, and it was a different world then, so they were very conservative,” explains Bobe, who joined the Harvard PGP in 2007 as both a subject and the project’s Director of Community. Not much was known at the time (and only a little more is understood now) about just what information a person could glean from a whole genome sequence, but it was clear that volunteers might learn, for instance, that they carried genetic variants predisposing them to serious diseases, or uncover sensitive details about their family histories. Participants would also run the risk of having their personal data accessed by third parties.
Bobe, whose task was to engage more volunteers in the study, acknowledged that there were real risks involved in equitable data sharing, not all of them known. Participants had to understand the basics not only of genetics, but also of research ethics and data security, to give their informed consent for the project. Still, he thought that requiring every member to have an academic background in genetics was far too onerous and, in his words, “undemocratic.”
“What we came up with was this entrance exam,” he says, “where people didn’t have to have a fancy degree, but could show that they were informed by passing a quiz.” This exam, originally around 50 questions long and requiring a perfect score to pass, was one of several innovations the Harvard PGP team would come up with during the rocky process of making members of their study into active participants in research, with full rights to own, share, and even publicize their data.
Today, the Harvard PGP has over 4,000 members and is being joined by parallel projects around the world, and lessons from the project can inform even broader efforts like Open Humans. The quiz format for informed consent, for example, has been baked into the Open Humans network from the beginning: any member who wants to make a dataset fully public — as opposed to sharing it one-on-one — has to pass an exam modeled on the one created for the PGP.
Methods like this, which respect participants’ autonomy but also come out in the open about risks, make it easier to win IRB support for new studies that want to practice equitable data sharing. For Bobe, this governance practice is not just an important tool for scientists who want to collaborate over their data. It’s something approaching a moral principle, that study participants should not be kept in the dark when they generate personal data for research.
“We really want to give oxygen to researchers who are trying to practice this new form of governance,” he says. “This is something that’s already been adopted in the medical world, where you have a federally protected right to request and get access to your medical records… There is no such right to access in the research community.”
Open Humans has emerged just as new technologies are making it easier for people to sign up for research and gather valuable data in their own homes, through their computers, smart phones, and wearable devices. Apple’s HealthKit and ResearchKit platforms, for example, are allowing researchers to design studies based on patient-focused apps that track the real time activities of participants. For future partnerships, PersonalGenomes.org is looking beyond academic researchers, to commercial partners whose products could provide new sources of data for projects conducted through Open Humans.
“As you’ve seen with Apple ResearchKit, researchers are getting more engaged with these tools and technologies, and incorporating those in their research studies,” says Bobe. His organization is already working with one potential partner, RunKeeper, whose widely used activity-tracking app has been adopted for fitness studies. The plan is to allow Open Humans members to import their own RunKeeper data directly into their profiles, rather than having to funnel it through individual research projects before it can be shared within the network.
Commercial partnerships like this could also be a convenience to researchers, who are usually more interested in analyzing their data than building ways to capture, transfer, and share it with their study populations. “At this point we want to be a partner to researchers, and help them with the technology side,” says Bobe, a process that will include creating an interface for researchers to return data to their volunteers.
Not all collaborations on the commercial side will go smoothly. Open Humans originally planned to launch with an integration with 23andMe, the Silicon Valley-based company that sells genetic tests to customers and uses the resulting data to conduct health research. The collaboration would have been a huge boon for Open Humans, connecting the network to 23andMe’s very large and scientifically engaged customer base.
But the partnership fell through, Bobe says, when 23andMe decided it didn’t want to support automated pipelines for its customers to share their genetic data with third parties. 23andMe, which uses its customer data for drug discovery efforts both internally and with pharmaceutical partners, has a financial incentive to keep this information relatively private. (Reached for comment, a 23andMe spokesperson told Bio-IT World, “Any 23andMe customer can participate in Open Humans by downloading their raw data file and uploading it to the Open Humans website. We believe this multi-step approach helps ensure a more informed decision than the single-step approach originally proposed by Open Humans using the 23andMe API,” adding that the company considered this “a necessary safeguard for our customers’ privacy.”)
While industry may sometimes be reluctant to let scientific data become fully public, academic researchers often have hang-ups of their own. “We know we’re going to face many challenges in the research community as well,” Bobe says, “with people finding all kinds of reasons why they can’t or shouldn’t or won’t share data with individuals.” There are strong forces stacked against equal data access, which range from well-meaning concern for research volunteers who may misinterpret complex data, to a desire to scoop competing scientists on new discoveries and publications.
The risks of equitable sharing should not be underestimated, especially when people are given the power to make their data fully public. Bobe himself has an Open Humans account linked to his PGP data, and some brief online sleuthing reveals that he is nearsighted, suffers from asthma, and is an asymptomatic carrier for the genetic disease hemochromatosis. This information isn’t terribly sensitive (Bobe’s large glasses are a dead giveaway about his eyesight anyway), but the ease with which it can be found does illustrate why members of the public might be uneasy about opening up their data. In a more extreme case, a PGP participant with a serious health problem might have very real concerns about how sharing this information could affect her insurance policies.
Even citizen scientists who don’t want to make their information fully public, however, can use Open Humans to make their participation in research more meaningful. An individual’s catalogue of intestinal bacteria collected for American Gut may be interesting on its own, but it becomes even more powerful when researchers can link it to the same person’s genome sequence through the PGP. That principle applies whether volunteers choose to share their data only with specific individuals, or with the entire Internet.
The promise of linking datasets from separate, interrelated projects also gives scientists an incentive to try out equitable data sharing, knowing that the data they gather could become richer by plugging it into the Open Humans network. As the project progresses, PersonalGenomes.org will seek out new research partners to expand the wealth of information that could be contained in any single member’s profile. “It certainly would be nice to try to find more research studies that have more synergy together, where you could have more sophisticated hypotheses when you bring together these four different datasets,” says Bobe.
Even more significant than joining together multiple datasets is Open Humans’ goal of creating a central hub for equal access research, bringing new citizen scientists into the fold and connecting old ones with a wider range of studies. Research on human subjects is still overwhelmingly a top-down affair, with data and recruitment closely controlled by the research institutions in charge, but enthusiasts like Bobe see the model shifting to more open and volunteer-driven practices.
“I think these mobile tools and platforms will help make this something of a cultural activity, where it becomes a normal part of life to enroll in a few research studies every year,” he says. “I think it’s going to be the main way of doing science.”
It’s an optimistic vision, but working with members of the PGP, Bobe has seen plenty of enthusiasm for research when it offers participants something back. Less than a month after launch, Open Humans already contains over 1,000 profiles — some from members of its three partner projects, and many from members of the public who are just intrigued by the possibility of becoming more involved in science.
As more of these members start to post their data on the network, Open Humans will also conduct a little study of its own, asking participants whether sharing data has led to any benefits or harms in their lives. Open data in the Internet age is still a contentious issue, and as Bobe and his colleagues work to create a safe space for new models of research, they plan to learn from this first generation of data sharing pioneers. As technology and changing healthcare practices continue to expose more of our information to the World Wide Web, projects like Open Humans may provide an early picture of how an “open sourced” human can live and thrive.