The finding is likely to kick off another round of controversy about how genetic research mixes with potential social bias.

An epigenetic scan can predict with 70 percent accuracy whether a man is homosexual, according to research presented today at the annual meeting of the American Society of Human Genetics.

“To our knowledge, this is the first example of a predictive model for sexual orientation based on molecular markers,” said author Tuck Ngun, Ph.D., a postdoctoral researcher in the Center for Gender-Based Biology at the University of California, Los Angeles.

Dr. Robert Green, Ph.D., a Harvard University geneticist, called it an “intriguing finding.” The lead author of the study is Dr. Eric Vilain, Ph.D., who is a geneticist at UCLA.

Sexual orientation is widely thought to have a genetic component. Studies on the genetics of homosexuality have garnered a lot of media attention, but they have come up with conflicting results.

An early study pointed to chromosome Xq28, but a later study didn’t find that it had as strong a correlation with homosexuality as other genetic markers.

In other words, there’s a lot of interest in a gay gene but not much else.

“There is absolutely nothing that’s well accepted that defines a genetic basis of homosexuality,” Green said.

Ngun agrees that his research is far from conclusive. It suffers from the same shortcomings as other genetic work: It points to a correlation but not a cause or mechanism between a genetic pattern and a trait.

Read More: Scientists Find Gene Editing with CRISPR Hard to Resist »

Because genes are slippery in this way, Ngun looked at methylation, or the mechanism by which genes are turned on and off.

The idea was that if there were differences in methylation in gay/straight twins, that would signal an area where the genes might affect sexual orientation.

Ngun’s early data identified five regions of the genome that seemed relevant. They then fed data on those regions into a computer prediction model. The model guessed if the study participant whose DNA it was reading was homosexual. It was right seven times out of 10.

However, the sample size was small: 37 pairs of twins in which one was homosexual and the other was heterosexual, and 10 pairs in which both were homosexual. It’s not clear if the predictions would be as accurate if participants were not twins, for example.

But the research points to two genes that should be added to the list of those that may contribute to homosexuality in men. KIF1A, a gene that affects nerve formation, and CIITA, a gene that regulates a type of immune molecule called MFC.

The link to the immune system echoes other recent research suggesting that people seek mates who have immunity unlike their own. In a heterosexual match, mismatched immunities in the parents would give their offspring protection from a broader range of immune challenges.

We may be able to subconsciously detect a person’s immunity through smell.

“What this research says is that odor may be part of attraction. And if you ask most people, they’ll say ‘some people turn me off and some people turn me on’ and that’s related to odor,” Ngun said.

Read More: Moving Closer to Genetically Engineered ‘Designer Babies’ »

The epigenetic findings leave a lot of questions unanswered, but a 70 percent prediction rate calls for serious reflection about what it would mean if we could point conclusively to a genetic signature of homosexuality.

“This is sort of one of the scenarios that everyone has feared — that there could be some spurious or even statistically valid link between a genetic marker and a trait that some people might consider undesirable — homosexuality, criminality, pick any trait you want — and that then becomes an opportunity for prejudiced people to oversimplify and express prejudice and discrimination,” Green said.

Ngun explained that as a gay man himself, he wanted to learn more about what makes him and others the way they are. But he has had second thoughts about the power this kind of research could have in the wrong hands. He is abandoning genetic research.

“It didn’t fit comfortably for me anymore,” he said.

While Ngun’s findings sound scary, they don’t make it possible to weed homosexuality from the gene pool. Genes turn on and off over time, and it’s not clear when the patterns of methylation that the predictions were based on emerged. They would be unlikely to occur before an embryo was implanted at a fertility clinic, he said.

But the use of genetic research to defend homosexuality also sits wrong with Ngun.

“While that’s clearly been a very effective argument — we’ve seen incredible advances — it’s not without its problems,” he said.

A better argument would be “to say whether it’s inborn, or no matter what the ultimate cause is, shouldn’t we be treating other human beings like human beings? We shouldn’t be saying ‘it’s not their fault.’”

Ngun also thinks we should be having a more robust conversation about the ethical use of genetic science to be ready in case a “landmark study comes along” pointing a neat red arrow to a gay gene. But for now, such a gene is more a metaphor than a reality.

Read More: What Are Sexual Norms? »