How Earnest Research Into Gay Genetics Went Wrong

In late spring 2017, Andrea Ganna approached his boss, Ben Neale, with a pitch: He wanted to investigate the genetics of sexuality. Neale hesitated. One of the top geneticists in the country, Neale and his colleagues at the Broad Institute, a pioneering biotech hub in Boston, had a decade earlier developed software that made it much easier for scientists to study the vast amounts of genetic data that were beginning to flood in. It was the kind of tool that helped illuminate a person’s risk of developing, say, heart disease or diabetes. And now, as Ganna was proposing, the approach could be applied to the foundations of behavior, personality, and other social traits that in the past had been difficult to study.

Ganna wanted to pounce on a new opportunity. A giant collection of carefully cataloged genomes, called the UK Biobank, was about to become available to researchers. A scientist could apply and then gain access to data from 500,000 British citizens—the largest public repository of DNA on the planet. To Ganna, the genetic origins of being gay or straight seemed like the kind of blockbuster question that might finally get an answer from a data set of this size.

Neale wasn’t so sure. As a gay man himself, he worried that such research could be misconstrued or wielded to advance hateful agendas. On the other hand, a better understanding of how genetics influences same-sex attraction could also help destigmatize it.

Then Ganna mentioned that another group was already pursuing the question using the UK Biobank: a geneticist named Brendan Zietsch, at the University of Queensland, and his colleagues. In 2008, Zietsch published a study suggesting that the genes straight people shared with their gay twins made them more successful at bedding heterosexual partners. Now he was going to further test this “fecundity hypothesis” with a much more powerful data set. He’d also proposed investigating the genetic associations between sexual orientation and mental health. Thinking his lab could add expertise coupled with caution to such a project, Neale agreed they should try to team up with Zietsch.

“Armed with the knowledge that this research was going to be done, I thought it was important that we try and do it in a way that was responsible and represented a variety of different perspectives,” he says, noting that, because there is so much genetic data to work with these days, collaborations in his field are commonplace “But it was also important to me personally, as a gay man, to get involved.”

From the outset, Neale expected some pushback and misunderstandings. That’s why he involved LGBTQ+ groups along the way, something not technically required for the kind of research he was doing. But he wasn’t prepared for scientists within his home institution to rise up and challenge the value and ethics of his work. And he was even less prepared for a company to exploit the results of the study—just a few weeks after it was published in the journal Science—to sell an app purporting to predict how attracted someone is to the same sex.

Sold as infotainment on a website called GenePlaza, the How Gay Are You? app might have seemed harmless to some. But it’s easy to envision how it could be abused, especially as more and more people get their DNA tested at younger and younger ages. Imagine the app getting passed around a middle school slumber party and stigmatizing results getting posted to social media. Or being used to screen applicants looking to rent an apartment or buy a condo (which would be perfectly legal under current US federal law and prohibited in only 22 states). In places with fewer civil rights protections, it’s even easier to see how such a tool could be wielded as an instrument of oppression. A state could surreptitiously collect DNA in order to persecute LGBTQ+ people, similar to how China has used genetic information to identify and imprison more than a million Uighurs, an ethnic Muslim minority in China. “It all depends on who’s using the data and why,” says Mildred Cho, a biomedical ethicist at Stanford who specializes in genetics. “Even if you don’t get sent to the gulag, there are lots of direct harms that can come out of ascribing labels to people based on their DNA.”

There’s no evidence the app actually harmed anyone before GenePlaza suspended it amid public backlash. But its brief existence made clear just how easily genetics research can get twisted and possibly weaponized in the era of enormous data. To bioethicists watching the saga unfold, it also exposed a weakness in the frameworks established to protect people from being harmed by research. Those systems revolve around research subjects—the people who consent to getting poked and prodded and experimented upon. But what happens when the people who are the potential victims of research are not the people who are giving the samples? Neale’s study is just one high-profile example of how the research ethics edifice—erected in a time before the internet, low-cost DNA sequencing, and genome-crunching AI—might not be ready for the age of subjectless science. And now that a global boom in biobanking is enabling researchers to probe the connections between people’s DNA and almost every facet of their lives, figuring out how to balance scientific freedom and progress with the risks of genetic discrimination has never been more urgent.

Soon after his conversation with Ganna, Neale reached out to Zietsch about joining forces to study the genetics of same-sex sexual behavior, and Zietsch agreed. A few weeks later, when the UK Biobank officially went live, they got their data. Through another collaboration, the team also gained access to a data set from 23andMe, which has been conducting its own research into human sexuality since 2012. They set to work combining the two to produce an even larger pool of genetic information.

If they had been working with human subjects, they would have had to pass through an ethics review. But because their data was stripped of identifying details and by definition subjectless, no such evaluation was required. Plus, the UK Biobank—which had obtained consent from the individuals who supplied the data—approved the project, finding it consistent with “health-related research that is in the public interest.” Still, Neale wanted to exercise caution, so he urged the group to take a few unusual steps. In July they registered their study with Open Science Framework, which would allow any member of the public to see what they were working on. They also convened two workshops in London through a group called Sense About Science to hear the perspectives and potential concerns of a handful of local LGBTQ+ advocacy groups. Ganna called in to share plans for the research and listen to feedback. Another team member, located in Cambridge, attended in person.

One concern that arose was why the research was focused solely on finding gay genes, as opposed to genetic links to sexuality more broadly. “We objected to the way such research has built-in bias and expressed reservations that the research could risk pathologizing same-sex relations,” says Pliny Soocoormanee, an LGBTQ+ advocate who works for the UK-based Peter Tatchell Foundation and attended a workshop. Sense About Science compiled the feedback into a report for the researchers. According to Neale, the main takeaway was how important it would be to clearly communicate their results with nuance and sensitivity to context. Given that the researchers already had the data they needed, the exercise was less about getting permission or LGBTQ+ buy-in for the research than about charting a responsible way to share its conclusions.

By the beginning of 2018, Neale’s team was beginning to get results. They looked at genetic variants—single-letter DNA changes—in people who said they’d had at least one same-sex partner and compared them to people who’d never had a same-sex encounter. This technique, known as genome-wide association, turned up five variants that stood out as being significantly associated with having same-sex sex, though the effect was small—less than 1 percent when combined together. The analysis also surfaced more common variants that contribute to between 8 and 25 percent of the variance in a population’s sexual behavior. In many ways, it confirmed older research in twin studies that being gay likely has a genetic component. But it also suggested that a single gay gene didn’t exist. More likely there were thousands, many not yet discovered, and that environment was always going to also play a role. When Neale’s team tried to use the genetic markers they’d found to predict how people in unrelated data sets reported their sexual behavior, they found almost no correlation. That meant that, at least for now, individual-level prediction is effectively impossible. In October, Neale and Ganna flew to San Diego to present these preliminary results at the American Society of Human Genetics annual meeting.

They knew their work would cause some uproar. To combat any misunderstandings, their group was busy building a website full of information about the study and its design. But they couldn’t publicize it until their paper came out in Science, to which they had already submitted their results, because of the journal’s embargo policy. So when Ganna took the stage at the human genetics conference, he mentioned the team’s ethics work but devoted the bulk of his 20 minutes to sharing his group’s findings.

At one point, he flashed a chart showing an overlap between genes associated with same-sex encounters and those linked to mental health conditions, including depression and schizophrenia. As the presentation went on, Joe Vitti, who was sitting in the crowd of more than 6,000 people, felt a growing unease. Also a postdoc at the Broad, and a member of its LGBTQ+ affinity group Out@Broad, he had no idea this work was taking place in the same building where he worked.

For too long, the medical community had classified being gay as a mental disorder. To see these traits presented together was disorientating, to say the least. Ganna did add some caveats: He was using data from mostly older British people, who had grown up in a deeply homophobic society where being gay was illegal. Some psychiatric conditions are more prevalent in queer communities—in 2017, 52 percent of LGBTQ+ people in the UK reported having experienced depression in the past year. Given the long history of discrimination, it struck Vitti as close to impossible to suggest genetics had in fact caused that mental suffering.

He immediately worried about how Ganna’s chart could be taken out of context and used in potentially harmful ways. He knew, for example, that genetic research was fueling fresh waves of white supremacist propaganda, with alt-righters chugging milk to promote their ability to digest lactose, a genetic trait more common in white people than others. Visions of Ganna’s graph cut and pasted into violently homophobic corners of the internet flashed before Vitti’s eyes. And so, as the talk wound down, he left his seat and strode across the room to join a line of people queuing up behind a microphone. Steeling his nerves, Vitti readied a question for Ganna. “Basically, I wanted him to assure me that they’d thought about what’s going to happen when this research gets out, that there’s good reason to believe this will help people and not hurt them,” Vitti says. But the moderator ended the session before he got to ask his question.

On his flight back to Boston a few days later, Vitti took out his frustration on his keyboard, typing out a summary of what had happened at the conference to share with the rest of Out@Broad. In it he outlined a long list of concerns. The most serious one, he wrote, was how the results of Ganna and Neale’s study might be misinterpreted and even abused when they were published. “Thinking on a larger scale, we can forecast opportunities for technological diagnostics of sexual orientation, or genetic embryo screening,” he wrote.

A few months later, in February, he and other Out@Broad members met with Neale to air their concerns about how the research had been conducted. (Ganna was in Europe and joined by phone.) They spent much of the hour discussing an early draft of the Science paper, which the researchers had provided to Out@Broad. Some suggestions—nix the rainbow-colored Manhattan plot (pandering), describe a single same-sex encounter as a behavior not a sexual orientation (imprecise, not supported by the data)—stuck. Others, like adding a note to the mental health correlation chart from Ganna’s presentation (now a figure in the manuscript) indicating there was no implied causation, were taken under consideration, though ultimately ignored.

Steven Reilly, one of the Out@Broad attendees, felt like the meeting came too belatedly to do much more. “We just went into harm reduction mode,” he says. “But it was too late to really address some of our bigger concerns.”

At one point though, he remembers trying to push Neale and Ganna to confront the possibility that the variants they had identified might be used in the future to further marginalize LGBTQ+ people, or to even pick embryos to erase future generations of gay people. He says they brushed the risks away. “The retort was always, ‘There’s no predictive power in these variants, it just won’t work.’ And to that I’d say, if someone in Chechnya, for example, wanted to use this test to cull gay people, and they kill 1,000 people and the 1,000 people they kill aren’t gay, that doesn’t really make me feel better,” Reilly says. (Chechnya is currently in the midst of an anti-LGBTQ purge, with more than 200 reported cases of torture and abuse at the hands of the state police.)

This was an extreme example, but Reilly had reason to worry. Just last year, a large consortium used UK Biobank data to publish the largest-ever study of the genetic influence on educational attainment. They wrote a 27-page FAQ explaining that they couldn’t forecast how long an individual might stay in school. And yet, shortly after, genetic tests claiming to predict academic achievement started popping up on DNA marketplaces like Helix and GenomeLink. (Helix no longer sells the app.) A company called Genomic Prediction also used the data to build a product it’s now selling to IVF clinics to screen embryos for low intelligence.

Neale hoped his team could keep their science from falling into the same trap. After all, they had brought on Robbee Wedow, a sociologist at Broad who had worked on the educational attainment study before joining the institute. Wedow, also gay, understood Neale’s project would need a much more involved outreach strategy to prevent history from repeating itself. He spearheaded many of the group’s efforts, including getting feedback on the paper’s language with a number of national LGBTQ+ organizations, including GLAAD. And on the day their study published in Science, the Broad Institute put out a press release which included links to critical essays written by Vitti, Reilly, and other Broad community members.

Neale admits they got some things wrong, including not engaging Out@Broad from the beginning. But he hoped that showcasing the diversity of perspectives would help provide some transparency. “We felt an obligation to try to represent the work in as honest and accurate a way as possible, to try to head off at the pass potential misuse to the best of our ability,” he says. “And, you know, I’m not sure there’s more we could have done.”

Nineteen days later, a rainbow-colored icon appeared on the genetic app store GenePlaza, with the name How Gay Are You? next to it. For $5.50 you could download the app, run your genetic data through it, and get a “same-sex attraction score.” If you didn’t already have a file of your genetic data from a company like 23andMe or Ancestry, GenePlaza would sell you a spit kit too.

An accompanying essay, written by the app’s developer, Joel Bellenson, began with: “A new study published in Science, on 29 August, and based on the genomes of nearly 500,000 people, has blown up the simplistic expectations of a single gene for same-sex attraction.” He went on to make some bizarre claims that perhaps the discrimination LGBTQ+ people suffer is “crucial for their fullest intellectual and social development,” comparing the creative prowess of “the Alphabet people” to the genius Jews are said to have acquired through enduring generations of oppression and genocide.

Bellenson was one of the first entrepreneurs to try to take advantage of the DNA age. He started a company called Double Twist that built some of the first genome-crunching software. In 1997, when Gattaca came out, Bellenson made the company’s 200 employees attend the Bay Area premiere. “I wanted them to understand that we are bringing enormous power into the world and we have to try to understand what it might mean,” he says. Double Twist didn’t survive the dotcom bubble burst, and Bellenson moved on to other things. (Another failed business venture of his, which sought to aromatize the internet with smell-enabled computers, graced the cover of WIRED in 1999.)

He now lives in Uganda, where he is involved with a local pharmaceutical company (and where same-sex relationships are currently illegal). Last year he was approached by Alain Coletta, a Belgian who had recently founded GenePlaza to be a marketplace where people could learn more about their DNA than what 23andMe or Ancestry would tell them. Pitched as infotainment, the site sells some ethically and scientifically dubious apps predicting, among other things, intelligence, math prowess, and depression. In the beginning, GenePlaza built most of these apps itself. But the company was looking to expand, and Coletta wanted Bellenson to help.

He says that, at first, he blew it off—until he read about Neale’s study. He told Coletta he wanted to make an app based on its findings. “I couldn’t think of any louder gong I could ring to say ‘Joel’s back.’”

When Vitti saw the app, he tweeted it out, tagging Neale and Ganna along with the Broad Institute. “I didn’t know whether to say it’s worse than we thought, or this is exactly what we told y’all was going to happen,” he says. He waited a few days, but Neale and Ganna didn’t respond. So Vitti, who by now had left the Broad and was working at genomics software company Seven Bridges, started an online petition to pressure GenePlaza to take down the app. Within a week, 1,400 people had signed it, as The New Scientist reported.

Neale says he and his coauthors were still discussing a course of action when Vitti launched his petition. He was tempted to ignore it, in the hopes it would go away. But as reporters began contacting Neale and his colleagues, they decided to write a letter to GenePlaza. Sent on October 14, it called the app a “gross and dangerous mischaracterization” of their work and urged the company to take it down.

At first, Coletta and Bellenson held firm. They informed Neale they had made some changes, including adding a second disclaimer that said “This app does NOT predict same sex attraction.” And they moved the majority of Bellenson’s essay behind the app’s paywall. But the app was still up a few days later, when Neale and Ganna flew to Houston for the 2019 meeting of the American Society of Human Genetics. On the opening day of the conference, news reports revealed that Ugandan politicians were moving to reintroduce a bill that would make gay sex punishable by death.

Again, Ganna took the stage. This time the topic wasn’t same-sex attraction but, rather, a newer genome-wide association project. Afterward, he took questions. The first one came from a human geneticist named Nancy Parmalee. She felt she hadn’t yet heard a satisfactory answer for why Ganna’s team had pursued the work given its potential to do harm. Parmalee argued that the creation of the GenePlaza app was an entirely predictable outcome that they hadn’t managed to prevent. “In light of this,” she asked, “do you still think it was ethical to publish the study?” A few people broke into applause. One man in the back of the room yelled out, “Not here!”

Ganna tried to deflect, saying he was there to talk about a different project. Parmalee pressed him, saying she felt that a forum of his peers was an appropriate place for him to respond to criticism of such a high profile paper. He declined to say anything more and the moderator moved on to the next question.

A few days later, GenePlaza quietly caved. On October 24, it changed the app’s name to “166 Shades of Grey” and removed the ability to purchase it, before scrubbing it from the site altogether, as Nature reported. According to Coletta, the impetus was the unjust portrayal of the product as a predictive and dangerous test with ties to Uganda. Coletta says that, in reality, the app was never sold to anyone in Uganda. “We would like to offer our most sincere apologies, as it was never our intention to offend anyone,” he added. “We clearly misjudged how sensitive this issue is.”

For now, Neale views this turn of events as a victory for science. But Tim Caulfield, a bioethicist and health law scholar at the University of Alberta, cautions that it’s too soon to start celebrating. “Research is very hard to control once it’s left your laboratory,” he says. He gives Neale and his collaborators credit for trying to be careful. The trouble is, it’s no longer their story to tell. “Pop culture owns this science now and it is going to keep getting spun to sell products or an ideology.”

Bellenson, for his part, is unrepentant. He has begun his own petition to reinstate his app on GenePlaza. And if that won’t work, he’ll go elsewhere. “I’m not a people pleaser. I do what I think is valuable,” he says. “If the tiny minority of drama queens want to scream about it, that’s their right, but I’m not going to stop doing what I do.”

All of which seems to put researchers like Neale in an impossible bind. If the data is out there, someone will do the science. And if the science is out there, someone will warp it to their own commercial or ideological agenda. Which is why bioethicists like Holly Lynch, at the University of Pennsylvania, think it’s time to reconsider whether existing ethical frameworks go far enough to protect people who aren’t the subjects of research themselves. “Historically we’ve considered deidentifed data to be low-risk—what’s the harm if we can’t figure out who’s in the study?” she says. But now, anybody who has had genetic testing done can have their sequence analyzed against markers like the ones Neale’s team identified, in a way that might potentially cause them harm. There’s even a name for this emerging phenomenon: bystander risk. “This falls totally outside the regulatory systems we have. It really is a new problem that we haven’t given enough attention to.”

Lynch says solutions will likely have to come from individual institutions. They could include additional levels of oversight for sensitive genetic studies involving anonymized data, such as stakeholder advisory boards, or requiring a bioethicist to embed on such projects.

Some researchers at the Broad are using this whole episode to push for similar types of reforms at their own institution. Members of Out@Broad are now leading an internal effort to evaluate additional rounds of ethical oversight for any big data genetics projects that involve vulnerable populations. Wedow, for his part, supports the effort. “We did everything by the book from an ethical perspective,” he says. “We crossed all our T’s and dotted all our I’s, but that doesn’t mean the rules always have to stay the same.”

Neale also supports taking a broader and more systematic view of how scientists think about research subjects. He knows from personal experience how sparse the existing guidelines are for working with anonymized genetic data. “How we use results like these should not be decided unilaterally by scientists. Those conversations are important and need to happen at societal levels.” But he also worries about how such efforts could cross a line into scientific censorship. “We should not mix up trying to understand and describe the world to the best of our ability with questions of what we should and should not do to people.”

Ganna, the man who kicked off the furor, now in charge of his own lab at the University of Helsinki, also seems ready to finally answer the question Vitti and then Parmalee posed to him. Yes, he believes they performed the work ethically. But like Wedow and Neale, he’s open to change. “Research ethics ought not to be viewed as static things,” he says. In fact, he’s planning to organize a session devoted to the topic at next year’s meeting of the American Society of Human Genetics. To his critics, that’s two years too late. But maybe, in this case, it really is better late than never.

More Great WIRED Stories

Read More