Now we’ve dispensed with the nonsense “gay gene” trope, let’s interrogate the way we talk about genetics and traits full stop

A recent study shows there is no one gene that determines homosexuality, but confusion still lingers on the predictive power of our genetic constitution. Far more interrogation is needed before the discussion reflects the reality of what genes "do"

September 04, 2019
Photo: Daniel Leal-Olivas/PA Wire/PA Images
Photo: Daniel Leal-Olivas/PA Wire/PA Images

Some people once felt that the reason to look for a “gay gene” was so that it could be eliminated. Under the headline “Abortion hope after ‘gay genes’,” in 1993 the Daily Mail wrote “Isolation of the genes [causing homosexuality] means it could soon be possible to predict whether a baby will be gay and give the mother the option of an abortion.”

Revisiting that reporting now it has been announced that there is no “gay gene” shows us how much has changed over less than three decades in public attitudes to homosexuality. Sadly, there has not been a comparable change in public attitudes to genes. Few would now take the abhorrent view that a genetic propensity to being gay would constitute good grounds for termination of a pregnancy, but probably many would have no problem believing that your genes could “make you gay” in the first place. This latest interrogation of genetic determinism illustrates how far we have still to go to shift the prevailing narrative about genes.

The new research, led by a team at the Broad Institute of Harvard University and the Massachusetts Institute of Technology, combed the genomes of nearly half a million people, using data from the UK’s Biobank repository of biomedical data, supplemented by that provided (with consent) by the American gene-sequencing company 23andMe. The researchers looked for correlations between genetic profiles and sexual orientation as reported by participants in the survey.

What they found takes some unpacking. First, there are some common features in the genes of people who self-report as being gay or attracted to the same sex: the researchers estimate that as much as 25 per cent of sexual behaviour could be explained by genetics. Second—this might at first seem to contradict that claim—one can’t, on the basis of these results, make any meaningful prediction about someone’s sexual orientation from their genome.

To unravel what all this means, we should first recognise that most if not all of our behavioural traits are influenced by our genes to some degree. It has long been known that the probability of a person being gay has an inherited component. In fact, comparisons of people’s genetic profiles have shown that there are statistical similarities in the genes of people who get divorced, watch a lot of television, engage in criminal activity—you name it.

At face value this seems absurd. The distribution of genes in a population is shaped by evolutionary pressures—that’s to say, by Darwinian selection (as well as random chance). Yet genes for “watching television” could not possibly have been selected (for or against) in the time since TV was invented. But the gene variants that are highly represented in people who watch a lot of TV are not in any sense “for” that purpose. These may be genes that influence more general, long-standing behavioural factors such as attentiveness, which may be manifested in their watching habits as well as in many other aspects of their conduct.

There’s an important clarification here: “genes” are often used as synonyms for “gene variants.” We all have essentially the same genes—that is why there is such a concept as the human genome at all. But we have different versions of these genes, which exert slightly different effects. Of the 16 or so genes that affect eye colour, for example, one version of the gene called HERC2 is partly responsible for blue eyes. These different versions are more strictly called alleles, and they differ in the precise chemical structure of the corresponding stretch of DNA, generally in some very minor way that tweaks the gene’s effect on how our bodies grow or work.

If there are genes that lead to blue or green eyes, then why not to same-sex attraction? One problem with the idea is that behavioural traits, unlike characteristics such as eye, hair or skin colour, are mediated by the brain. And how the brain gets wired up—how the 100 trillion connections are made between its 100bn neurons—is not wholly specified by our genes, since we don’t have anywhere near enough genes to dictate every connection. There are chance elements at play during development that can affect sexual preference, making it (partly) innate without being genetically prescribed.

The deeper problem with supposing that there might be a gay gene, though, is that it has been obvious for many years now that this is not, in general, the kind of thing genes do. They do not usually link in direct, one-to-one fashion with an organism's set of traits (what biologists call the phenotype). There is a strong inherited (and thus genetic) component to height, but you can’t point to a gene, or even a few genes, that control it. The fact that we have five fingers is genetically determined, but there isn’t a gene for five-fingerness.

So what are the genes doing to put us together? The answer is extremely complicated, but we can say for sure that many of the key genes involved in development don’t have a “function” one can define in a sentence with reference to the phenotype. The effect they have on the developing organism may depend on where and when they get switched on or off. This control of gene activity is influenced by a host of factors—in particular, by other genes. They operate in networks, and some important genes have influences that depend on the context in which they are activated.

What all this means is that we lack a language—we have no good metaphors—for talking about what the role of genes is. The problems arise because we thought we did. When the molecular basis of genetics—in particular the way parts of DNA encode instructions for making proteins—became apparent in the 1950s and 60s, it was thought that genes could be regarded as instructions for making the body. This was the beginning of the notion of genes “for” this or that feature or trait: genes for height, eye colour, sexual orientation.

Most geneticists now deplore this idea—but it was sowed by their fellow researchers, past and present. Even now there seems rather little concern to find better metaphors: some geneticists remain content to talk of the genome as the “instruction book” or even the “blueprint” of the human body. The difficulty, now painfully apparent precisely because we have so much genetic data to draw on, of linking genes to traits (such as common diseases) is typically dismissed as an indication that “biology is complicated,” rather than an admission that biology doesn’t work in quite the way we once thought.

“The fact that we have five fingers is genetically determined, but there isn’t a gene for five-fingerness"
The recent advent of so-called genome-wide association studies (GWASs), which trawl through genetic data banks looking for gene variants statistically linked to traits, has both helped and hindered. Rather than seeing if specific alleles are associated with a trait, GWASs can look without prejudice through the entire genome (our full complement of genes) and identify any correlations, no matter how small. For a complex trait such as intelligence, GWASs have identified hundreds or even thousands of gene variants that have some influence. In general, the effects of individual genes are negligible—but collectively they might explain a significant fraction of the variations in that trait found between individuals. Many tiny effects are typically summed into a single “polygenic score” that can be calculated for any individual and used to predict traits.

The Broad Institute work (which involved researchers from many other places too) took this GWAS approach. Despite casting the widest net to date, it is still an imperfect analysis, because the data comes from a relatively older group (40 to 69 years old), of European ancestry and of relatively high socioeconomic status on average. The survey also relied on self-reporting of sexuality (fewer of the older people reported same-sex relationships) and necessarily had to impose a rather simplistic yes/no categorisation on sexual behaviour.

A few of the gene variants netted in the analysis seem to correlate rather more strongly than the others with same-sex attraction, although it isn’t really known what they “do.” One is linked to male-pattern baldness—which is influenced by sex hormones, so this might make sense in a hand-waving way. One might be associated with smell, which is thought to have a role in same-sex attraction. But there is certainly no “gay allele” of even these genes: it is not possible to predict whether people with these specific alleles will be gay or not. Taken together, the polygenic score for the five most prominent variants in the study could predict less than 1 per cent of sexual orientation—which is to say, they are not predictive at all. The discrepancy between this number and the estimated 8-25 per cent heritability of sexual orientation suggests that there may be many other gene variants that have tiny effects, which are yet to be identified. If so, it’s likely that those genes too will influence traits that have nothing to do with sexuality per se—for example, propensity to take risks, which might in particular social contexts influence whether a person will acknowledge or express gay behaviour.

This research has been conducted and reported by the Broad Institute team with commendable responsibility. They created a website to explain the nuances of the findings, and consulted with LGBTQ advocacy groups. All the same, geneticist Joseph Vitti, a researcher at the Broad Institute who was not involved in the work and identifies as queer, says that he struggles to understand the motivations for the study. “I have yet to see a compelling argument that the potential benefits of this study outweigh its potential harms,” he writes. “My strong intuition is that this study will be most enthusiastically touted by individuals unwilling to engage with the subtleties and limitations of its interpretation.” (As one tweeter put it, why not instead sequence the genomes of those prejudiced against gay people to find out what their problem is?)

The authors of the study argue that their work shows the complexity of human sexual behaviour. But it’s not enough to suggest that it also shows the complexity of how genes influence that behaviour. What is really at issue here is the notion of biological causation itself. Saying that genes even partly cause homosexuality is not far removed from saying that there can be such a thing as a (slightly) gay cell. As well as killing off the “gay gene” trope, it would be good if this study lays to rest the dangerous idea that our genomes are a kind of “crystal ball” that predicts who we are—and if it helps motivate a search for better ways of talking about what genes do.