The notion of a “gay gene”—a genetic basis for same-sex attraction—has preoccupied scientists, gay civilians, and folks who’d rather queers didn’t exist for decades. The discovery of such a mechanism would have profound implications, as I’ve explored previously for Slate: On the one hand, we’d know more about how human sexual diversity has evolved over our history as a species. On the other, the same knowledge could be weaponized in embryonic screening by bigots seeking to prevent future queers from being born. Until recently, all of this was largely theoretical. But today, we took an important step toward real-world gay genomics: The results of the largest-ever project designed to harness modern genome sequencing to understand the origins of human sexual orientation have been published, and it’s a remarkable document—illuminating, admirable, and troubling in turn.
The paper, published in the journal Science by an international team led by the Broad Institute’s Andrea Ganna, reports analysis of genetic data and self-reported sexual behaviors from about a half-million participants in the U.K. Biobank and customers of the personal genetics company 23andMe. Purely in terms of a search for “gay genes,” it is kind of a flop, though it does shed new light on the biology and evolution of human sexual diversity (more on that shortly). Of equal, if not deeper, interest, however, is how Ganna and his co-authors handle and present scientific work that may have far-reaching impacts on a marginalized community. The authors consulted with queer advocacy groups in presenting their results, and it really shows. The paper has a prominent sidebar devoted to clarifying that its focus is same-sex sexual behavior rather than queer identity, and explains other methodological and terminology choices. It also makes a point to say that the term “nonheterosexual” is not meant to be pejorative, which would honestly never have occurred to me. Yet for all this care, I’m not at all sure they’ve resolved the risks inherent in this line of research.
To their credit, the authors are open about the myriad challenges of the project. Same-sex sexual behavior is difficult to measure, because in a disapproving social context, people may never act on same-sex attraction, or may not disclose same-sex experiences to researchers. The first figure of the paper demonstrates this: Older participants were substantially less likely to report same-sex activity than younger ones. The authors address this issue, to some degree, by looking first for genes associated with whether participants reported any same-sex experience; then, using only data from those who reported at least one same-sex experience, looking for genes associated with having a greater or lesser proportion of same-sex partners. They also analyze men and women as separate cohorts—just including women in the study, much less treating their sexual orientation as a co-equal focus of the project, is an improvement over some prior studies.
The primary result of all this is perhaps underwhelming, at least if you were hoping for microscope images of a rainbow-colored segment of DNA. As big as it is, the data set isn’t enough to pinpoint individual bits of genetic code, or loci, that explain a substantial fraction of variation in the sexual histories of study participants. This is a long-established problem in genetics. Even if a trait has a genetic basis—as seen in earlier studies of sexual orientation across family trees or in pairs of twins—when that trait is affected by many loci, the contribution of any one locus can be too small to detect. In this case, only five loci had effects big enough to identify confidently: two associated with same-sex behavior in men; one in women; and two more showing associations in both men and women.
Even these loci have very, very small effects. At the locus with the strongest association to same-sex behavior, a man carrying what we might call the “gay variant” has a 10 percent greater chance of having had sex with other men. If the baseline probability of this is about 4 percent (the frequency of gay men in the U.S. population), that means that out of 1,000 random guys on the street who carry that variant, 44 have actually had sex with other men. For comparison, something like 154 out of 1,000 men living in San Francisco identify as gay or bisexual—which means simply knowing a guy who lives in that ZIP code is as much as three times more indicative that he might be gay than knowing he carries the right variant at this one locus.
This result is consistent with the history of failure to find genes that are robustly associated with sexual orientation; if orientation were determined by a few loci with strong effects, they’d probably have showed up in earlier, smaller studies. It’s also consistent with the most pedestrian hypothesis for the evolutionary persistence of people who are disinclined to have the kind of sex that makes babies: Although natural selection should act against genetic variants that contribute to same-sex attraction, the contributing variants have such small effects that selection against any one of them is correspondingly weak. Variants that could contribute to same-sex attraction stick around because they’re carried in smaller-than-critical numbers by many folks who are mostly or exclusively attracted to the opposite sex. Queerness happens by a kind of rare and fortuitous genetic convergence, when enough of the right variants come together in a single person.
The failure to find strongly predictive loci also neatly evades the most worrying prospect of this research: that its results could be applied in a genetic test for sexual orientation. Such a test could, in principle, be used to screen embryos or as evidence against adults accused of homosexual activity in a country where it’s illegal. Ganna et al. address that concern head-on by piloting a genetic test right in the paper. Pairing their results with genetic data from several earlier, smaller studies that collected sexual behavior data, they calculated “polygenic scores” for the participants in those prior studies—essentially, composite genetic estimates of how likely those people were to have had same-sex sexual experiences. Those polygenic scores were found to be correlated with the sexual behavior reported by participants in the prior studies, but there’s so much statistical scatter around that correlation that the scores “do not allow meaningful prediction of an individual’s sexual behavior,” as the authors say in the paper’s abstract. It is, I think, the first time I’ve seen a genetic study highlight a lack of predictive power as a primary result.
Ultimately, this paper represents to my mind the best possible way that a modern study of the genetics underlying sexual orientation could have turned out. It was informed by the community most affected by its results, designed to account for at least some of the complexity of sexual orientation (that identity vs. behavior distinction), and proved unable to produce the basis for a reliable genetic test of sexual orientation. That said, the hazards of this line of research aren’t really mitigated by ethical research practices, in the conventional sense. Those hazards lie not just in how the work is conducted and presented, but in how society receives it.
There is, as of now, not much regulation of how genetic data is used once it’s published. Anyone who pays for a personal genomic read from 23andMe can look up whether it registers variants associated with same-sex behavior in the new analysis, regardless of the actual predictive value of those variants. Lack of demonstrated predictive power hasn’t prevented the emergence of startups offering to tailor diet advice to your genetic profile, and even proposing to screen embryos for future traits based on polygenic scoring methods.
And there’s an established cottage industry of quackery purporting to “cure” queer people—though campaigns to restrict and ban such conversion therapy are making progress, there are plenty of U.S. states and jurisdictions across the world where it’s not meaningfully regulated. Will we see conversion therapists drawing on results from this paper, maybe trying to preempt homosexuality in children too young to even articulate a sexual orientation? As much as Ganna and his coauthors have done to try to head off this kind of misuse, it’s now beyond their control.
There’s also very little to prevent follow-up research by less cautious geneticists. For the foreseeable future, it will only get easier to collect larger data sets than Ganna et al. were able to assemble—and bigger data sets will be able to resolve loci with smaller individual effects. That could get us closer to an effective genetic test for orientation, or reveal enough of the underlying biology to suggest pharmaceutical treatments to change same-sex attraction. Scientists, as a group, don’t like to leave research questions unanswered.
I’ve been gratified that my previous writing about the concerns raised by genetic studies of sexual orientation has seen a generally supportive reception from other biologists. (At least, the ones who’ve talked to me about it.) To the extent that anyone’s raised a counterpoint, it’s more or less this: How can we reasonably foreclose a line of research, even if its possible outcomes are dangerous? In response, I’d actually say that I don’t expect us to. My feeling, in that earlier essay, was not so much that this scientific work could be stopped, but that it’s coming to fruition whether we’re prepared for it or not, and before society is ready to receive it in a way that doesn’t put queer folks at risk. I wish I could read this new study purely as an answer to my own curiosity about the biology and evolution of my queer identity. I hope, in the end, that’s all it turns out to be. But, reading the new paper, I know I’m not the only one with these worries—it looks like, on some level, the people actually doing the research are worried, too.