Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.


Retiring the Single Gay Gene Hypothesis

The once-prevailing concept of a sole “gay gene” dictating sexual orientation has been put to rest in a powerhouse study published today in Science. The work illustrates the nature of science: evolving with the input of new data, especially the large-scale contributions of bioinformatics and crowd-sourcing.

“We formed a large international consortium and collected data for more than 500,000 people, comparing DNA and self-reported sexual behavior. This is approximately 100 times bigger than any previous study on this topic,” said lead author Andrea Ganna, of the Institute of Molecular Medicine in Finland and an instructor at Massachusetts General Hospital and Harvard Medical School, opening a news conference earlier this week.

Human Sexuality is Nuanced and Complex

The investigation lowers the estimate of the genetic contribution to same-sex sexual behavior, thanks to analysis of a trove of data from the UK Biobank and the consumer genetic testing company 23andme.

I hope that the demonstration of a diminished role for genetics will counter the idea that having sex with a person of the same sex is something biologically broken that needs to be fixed. “Using these results for prediction, intervention, or a supposed ‘cure’ is wholly and unreservedly impossible,” points out Melinda Mills, a sociologist from the University of Oxford in an accompanying Perspective.

Chromosomes and genes tell only part of the story of human sexuality.

Sums up team member Ben Neale, from the Broad Institute and the Analytic and Translational Genetics Unit at Massachusetts General Hospital:

“This study provides the clearest glimpse yet into the genetic underpinnings of same-sex sexual behavior. It highlights the complexity of the genetics, but genetics is not the whole story. It also underscores an important role for the environment in shaping human sexual behavior and perhaps, most importantly, that there is no single gay gene, but rather the contribution of many small genetic effects scattered across the genome.”

Thousands of genes each exert tiny influences.

The international research team developed a website (password gwas) about the study that uses plain language, after consulting LGBTQIA+ outreach and advocacy groups.

That input was critical, considering the murky history of the search for a “gay gene.”

In 1993 Dean Hamer, then at the NIH, famously published a paper in Science hypothesizing that a gene or genes predisposing a man to same-sex behavior resides near the tip of the long arm of the X chromosome, a region dubbed Xq28. The idea stemmed from the observation that gay men tended to have more gay relatives on their mothers’ side than on their fathers’, suggesting a role for the X, which males get from their mothers. The evidence: 33 of 40 pairs of gay brothers share that region.

Through the lens of today’s big data, it’s amazing that such a limited study could have had such traction, but it inspired headlines back then. Here’s how John Crewdson, writing in the Chicago Tribune in 1995, attempted to temper the hype: “Dean Hamer doesn’t claim to have found the gene, or even the set of genes, that contributes to a propensity for homosexuality. What Hamer reported in a July 1993 article in Science magazine is statistical evidence that such genes exist.”

It isn’t that simple.

(image credit Francesca Cattaneo)

Fah Sathirapongsasuti, senior scientist for computational biology at 23andMe and a team member, remembers reading the Hamer paper as a teen trying to understand his sexuality:

“I looked at the Internet for “gay gene” and came across Xq28, and readily blamed it on my mom because she gave me my X chromosome. Well, luckily for her, these papers have basically disproved that theory and now the blame is shared equally between my mom and my dad.”

The turnaround celebrates how science works. We disprove hypotheses in science, we don’t ever “prove” anything – the advertising industry invented “scientific proof.” Conclusions can change as new data accumulate – that’s especially likely when technology accelerates, such as replacing 40 data points with more than half a million, as is the case in the new work.

Comparing Points Along Genomes

The new study is a “GWAS,” one of the worst abbreviations in genetics, which has quite a few. GWAS means “genome-wide association study.” It compares a large set (at least hundred of thousands) of single-base sites (SNPs) in genomes that vary – that is, can be A, C, T, or G – among individuals.

If people with a specific trait or illness share sweeping GWAS patterns, then a genetic component is inferred. It’s a little like comparing many points of interest encountered on a cross-country drive and concluding, if two cars share the same stops, that they’ve followed the same route.

With high numbers of SNPs and people, powerful associations can emerge. Still, validation in other populations, plus deep knowledge of biochemistry and cell biology, are required to elevate an association or even a correlation (a directional association) to causality.

A GWAS is sometimes simplified to a “polygenic risk score,” but either way the measurement is a subtle look at genetic variation. It’s not an analysis of a single mutant gene, like the one behind sickle cell disease.

The UK Biobank contributed responses from 408,995 people and 23andMe contributed 68,527. But it’s the nuanced nature of how the company amassed its data that lies behind the revelation that is really an affirmation of the many ways that people express their sexuality. 23andMe allowed for fluidity of sexual preference.

23andMe: Beyond Binary

Participating in 23andMe research is fun. Their emails are piling up in my in box right now:

“Just a friendly reminder that your:

Matching speed cognitive game

Healthy aging survey

Spot-the-difference cognitive game

Hearing-in-a-crowd test

Recent dietary habits survey

is available in your 23andMe account.”

The “sexual orientation” survey is popular, Neale said.

Like many other geneticists, I didn’t at first support what was initially called direct-to-consumer genetic testing, but I changed my mind after the company identified two genes behind Parkinson’s disease in record time, thanks to customers sharing data.

For the new study, 23andMe asked:

“With whom have you had sex?” and for people reporting ever having had sex with the same sex, followed up with a fleshing out of the answer: “other sex mostly,” “other sex slightly,” “equal,” “same sex slightly,” “same sex mostly,” or “same sex only.” If the UK Biobank approach is like a true-false question on an exam, then the 23andMe approach is like a detailed multiple choice question. Nuance.

After the math, for a GWAS is a computational tool, the researchers concluded that whether a person has ever or never had sex with the same sex is distinct, in terms of the minuscule genetic influence, from the proportion of same-sex partners. Therefore, the two questions “capture aspects of sexuality that are distinct on the genetic level, which in turn suggests that there is no single continuum from opposite-sex to same-sex sexual behavior,” the researchers write.

This complexity is in stark contrast to the Kinsey scale, commonly used to measure sexual orientation, which assumes that greater attraction to the same sex diminishes attraction to the opposite sex. “The Kinsey scale is an oversimplification of the diversity of sexual behavior in humans,” said Neale.

The Data

The GWAS showed that the genetic contribution to variation in same-sex sexual behavior is 8 to 25%. It assesses contribution to the variation in the trait, not to the trait itself. “The effects are so small that this genetic score cannot in any way be used to predict same-sex sexual behavior of an individual,” Mills writes.

That’s critical. We can’t look at a person’s DNA sequence and predict sexual behavior.

Five SNPs stood out as being “significantly associated with same-sex behavior” compared to the thousands of others, two in males only and one in females only. Finding these five, though, is like looking at a very tiny pebble in a handful of sand of many particle sizes. Overall females and males share only 63% of the gene regions associated with sexual orientation, which is less alike than for many other traits. Not terribly surprising.

One of the SNPs is near genes that affect the sense of smell (for males). Two others are near genes that regulate sex hormones, in females and males. And none were on the X chromosome, for so many years thought to house a single gay gene.

The researchers identified several characteristics exhibiting a “genetic correlation” to same-sex orientation:

  • Personality traits of loneliness and openness to experiences
  • A fondness for weed and cigarettes
  • Depression (females more than males)
  • Schizophrenia (females slightly more than males)

Alan Turing

But no physical traits went along with same-sex experience, perhaps countering some stereotypes.

The researchers point out that the apparent increase in mental illness can just as easily reflect the effects of prejudice as it does DNA, especially among older participants from the UK who recall the criminalization of same-sex behavior. Cryptanalyst Alan Turing is a telling case. He was convicted of “gross indecency” in 1952 and given a choice of prison or taking estrogen. He chose the hormone and killed himself two years later.

The genetic input leaves room for other influences. “Sexuality is dynamic, with the ability to express and realize sexual preferences, and is thus also shaped and regulated by cultural, political, social, legal, and religious structures,” Mills writes.

The new data seem to echo what people already know from just being people. “The diversity that we see in the genetic analysis reinforces the message that the expanding acronyms in the LGBTQIA+ community is real,” said Neale.

A One-of-a-Kind Report

The Eunice Kennedy Shriver National Institute of Child Health and Human Development funded the research to probe the genetic influence on sexual orientation. The project began in 2012.

The Science paper is remarkable in the interactions with the community of people the research addresses. A special box explains word choices. For example, instead of “homosexual,” the term used is “nonheterosexual,” meant only to distinguish clearly, not to judge. Female and male are the biological terms used; not woman and man.

The researchers mention the limitations of their study: reliance on older, Euro-centric populations, as well as the bias in the 23andMe data. In various societies, the percent of people having sex with the same sex ranges from 2 to 10%, yet 19% of the 23andMe respondents report ever having sex with the same sex.

Because the team took such extraordinary measures to relay their important message clearly, I’ll end with their interpretation:

“Our results overwhelmingly point toward the richness and diversity of human sexuality. … not toward a role for discrimination on the basis of sexual identity or attraction, nor do our results make any conclusive statements about the degree to which ‘nature’ and ‘nurture’ influence sexual preference.”

  1. The “gay gene” was a straw man invented by anti-gay right wing fundamentalists. I’m sure it wasn’t your intention, but your article plays right into their hands. It’s the same phenomenon that occurs when good-minded people criticize Trump’s talk of an “immigrant invasion;” it ends up making the idea, no matter how ridiculous, more real. OUT Magazine had a much more fitting headline: There is No Gay Gene – There Are Thousands

  2. Thanks for sharing your opinion Chris, but as a geneticist, I’d have to disagree. The alternate title you mention, that there are thousands of gay genes, is just wrong, scientifically speaking. The study was a GWAS – a look at signposts, not actual protein-encoding genes at all. That’s clear in the full paper. The news conference and press release do not explain the science to this depth, they oversimplify. The signposts are near genes that might have something to do with sexual orientation. Those genes do plenty of other things — they might be involved in estrogen or testosterone synthesis, for example. So each of the thousands of genes has some sort of role in biology, and that role can affect sexual orientation. To call them “gay genes” is not accurate. I realize I may have made errors in my word choices that relate to sexual orientation because I do not know much about that, but I do know the science quite well. I don’t mean to be obnoxious, but I could tell from some of the questions at the news conference that some reporters didn’t have a clue about the science, comparing the GWAS findings to single-gene diseases. One of the researchers actually made a comment that the level was too high to even talk about. So, I did my best. Also, editors usually write the headlines. The OUT reporter might not have even meant that there are thousands of genes. These researchers did an exceptional job of explaining what they did.

  3. The truth is the “gay gene” was an invention by homosexuals to try to justify their sexual choice. It is all part of their contention that they were “born that way” and are unable to be any different. This has now been proven false, though anybody with any impartiality or intelligence knew the truth anyway. It will not end here though, there will be further inventions to justify their position. Those with bias will continue to believe the lies as they want to, and those who are impartial will continue to see through them.

  4. The truth is the gay gene was an invention by homosexuals to justify their position, as they contend that they were “born that way”. This has now been proven false.

  5. Chris Lee
    Your pro homosexual comments — which I can care less about —- are biased in the same manner you deny a south of the border invasion. There also must be a GWAS for high denial, a GWAS that inverts black to being white and another GWAS that converts Up to Down. Perhaps Dr Lewis can identify these as well?!

  6. Ricki, finally a great comment on this paper in the maremagnum of irrelevant copy&paste news. I have a question for you. Knowing the possible association between the “homosexuality” with high risk decisions taking trait and unexplained high percentage of “non-pure-heterosexuals” in the 23&me cohort (19%). I know it is crazy, but wouldn’t it be possible that there is a high representation of liers in the cohort and that any of the SNPs found would be a risk factor for lie-telling personalities or other trait instead of homosexuality as the p values are so unconvincing? Also, do you have any possible easy explanation apart for a random association to explain that the PPFIA2 gene that is in the chr.12 region has been found to be upregulated in cannabis users (PMID: 30879928), been cannabis use one of the traits associated with “homosexuality” in this study?

  7. Ricki,

    finally a great comment on this paper in the maremagnum of irrelevant copy&paste news.

    I have a couple of questions for you, I would love to have your opinion.

    1. Knowing the possible association between the “homosexuality” with high risk decisions taking trait and unexplained high percentage of “non-pure-heterosexuals” in the 23&me cohort (19%). I know it is crazy, but wouldn’t it be possible that there is a high representation of liers in the cohort and that any of the SNPs found would be a risk factor for lie-telling personalities or other trait instead of homosexuality as the p values are so unconvincing?

    2. Also, do you have any possible easy explanation apart for a random association to explain that the PPFIA2 gene that is in the chr.12 region has been found to be upregulated in cannabis users (PMID: 30879928), been cannabis use one of the traits associated with “homosexuality” in this study?

    Thank you very much,


  8. Something that makes me skeptical is that they have not found anything in Xq28 (Hamer et al 1993) or in the 2 regions in chr12 (SLITRK6 gene) and chr13 (TSHR) reported by Sanders et al 2017 using GWAS. It seems that they never confirm any region found. Please, anybody give me your opinion.

  9. Hi Daniel. First, thank you for the word “maremagnum” — I’d never heard it. Yes, the paper raises questions and has some inherent biases, like the self-selection of the 23andMe cohort and especially as you point out the inability to tell whether people are telling the truth. I too wondered why the X-linked site didn’t come up. The comment below about inventing a GWAS for anything and everything is also insightful. I’ve never liked these studies, because I think it’s possible to alter the outcome and conclusions based on whom you ask to participate as well as the questions posed. The paper to me was quite confusing and much of the media got it wrong (or maybe I did). The Hamer study had a big effect and small sample size, then this current paper initially had %s up to 24 or 25, looking at 500,000 or so people. I think the statistics were extrapolated to the world population, thereby including very rare SNPs, and that’s what sent the % plunging to below 1%. If someone can explain this more clearly, please do. But the take-home message for me is that the data reveal that there are many ways to become who we are, in terms of sexuality. Lots of us already knew that. Thanks for writing!

  10. Man, sometimes it’s really hard to look past the title and actually read the article huh? But then again, ignorance goes best with laziness.

  11. Much as I appreciate your attempt to explain this study to the general public, your description of the results is incorrect and misleading. Specifically, your claim that “genes explained less than 1% of the variation” in same-sex sexual behavior is a flat-out misrepresentation of the findings from the study.

    In fact, the study found that AT LEAST 8-25% of the variance in male and female same-sex behavior was due to differences in DNA. These estimates were obtained by analyzing ALL of the single nucleotide polymorphisms that were assayed and pooling the effects. Actual heritability is probably larger since GWAS studies have consistently been shown to underestimate genetic influence, a phenomenon know as “missing heritability” ( Indeed, by analyzing the UK Biobank, which has familial as well as DNA data, the authors “estimated broad-sense heritability—the percentage of variation in a trait attributable to genetic variation—at 32.4%.”

    The 1% number that you cite was obtained by looking at only a FEW of the polymorphisms detected in the study. This analysis has nothing to do with heritability; it was a way to compare different databases to one another.

    Please correct your article accordingly. You might also want to stress that the phenotype looked at in this study – ever or never having a same-sex sexual experience – is highly heterogenous and would never pass muster in a sexuality journal. Hopefully future studies will use more appropriate, verified measures.

    One thing I’ve learned in 25 years of communicating to the public about human sexuality is that facts matter, and that untruths will be exploited by bigots – as you are already beginning to see in this comments section. I hope you will prove your good intentions by making the necessary revisions to your blog.

  12. Thank you Dr. Hamer for your reasoned and respectful response. You are certainly the expert on this subject, not me. I apologize if I’ve misunderstood the statistics, but I had thought that the 1% came from extrapolation to a larger, perhaps global, population. Rather than rewrite my post, I refer readers to Dr. Mills’ piece that accompanied the original report, wherein she explains the origin of the 1% figure Here is what I think is the relevant part: “When using a different technique called SNP-heritability—comparing the genetic similarity of all unrelated individuals in the sample with their phenotypic similarity of same-sex sexual behavior—Ganna et al. found that genetics could eventually account for an upper limit of 8 to 25% of same-sex sexual behavior of the population. However, when all of the SNPs they identified from the GWAS are considered together in a combined score, they explain less than 1%. Thus, although they did find particular genetic loci associated with same-sex behavior, when they combine the effects of these loci together into one comprehensive score, the effects are so small (under 1%) that this genetic score cannot in any way be used to predict same-sex sexual behavior of an individual.”

    Please don’t shoot the messenger.

  13. The problem is that you flat-out state that genes account for only 1% of variation, which is simply untrue. Otherwise, this untruth will be exploited by anti-LGBTQ fundamentalists and conversion-therapists at the expense of my community.

    The 1% figure you cite has nothing to do with “a larger, perhaps global population,” and I don’t understand why you might think that. There is nothing about that in the paper, and Mill’s article states the correct conclusion, which is that the combined score accounts for only a small proportion of the actual heritability that was seen in the study..

    I have no interest in “shooting the messenger.” I just want the message to be correct. The post needs to be corrected – not by a reference in the comments to another article, but in the post itself. Please act like a responsible journalist and fix it.

  14. Dean, is it reliable to base the 8-25% calculations on SNPs that do not reach statistical significance? Don’t you think that some of the results obtained are contamined by “high risk decision” heterosexuals. Don’t you think that if there is any genetic risk factor it would also be for many other behavior traits?

  15. Dean, I think that this is not a one direction highway. Human behavior is complex and multifactorial. I find intriguing that such a deterministic point of view has been hold by your community and has been tried to be imposed as a dogma. It is frankly hard to see good and solid scientific papers on the topic.

  16. Dean, wouldn’t a “a trait attributable to genetic variation—at 32.4%.” also may be partially due to non-genetic factors?

Leave a Reply

Your email address will not be published. Required fields are marked *

Add your ORCID here. (e.g. 0000-0002-7299-680X)

Back to top