Hacker News Reader

Comment by coding123 | Hacker News Reader

I'm curious if they instead focused on a simple neural net that is fed genes + gay/straight flag if the NN would actually successfully predict it.... in other words, NN's don't care about "finding" that specific gene, they kinda include everything and it "finds" the gene without actually pointing it out.

That's very different from the human approach which is to try to find something specific.

I mean compare it to blood diseases based on genetics. There are many genes that lead to health problems that comes down the shape. But it's not just one thing, it's a spread of things. We're thinking Gay is a specific gene we're just as unlikely to find it.

Not arguing any specific direction, but I just suspect the findings if they don't include variances in their methods.

View on HN

Neural nets for genetics are probably a good fit, and perhaps arguably, a better biological analogy than actual neurons. But they're likely already getting the gist here. They actually say that with genetic analysis (implying more than one gene at a time), they could explain 25% of sexual behavior. To be honest, that's a lot more than I would have expected. In biology, that's a pretty high predictive power, and I think is a way more interesting headline than "no single gene controls sexual orientation".

At the end of the day though, many other factors will contribute, so genetics can only explain so much. With aggressive data collection during development, I would wager we could get sexual orientation prediction much higher. Probably not high on the proverbial bio-ethicist's wishlist of things to make broadly available.

For complex traits, I believe people think that linear effects dominate. Which is another way of saying that they see no advantage to adding a hidden layer, which encodes interactions, compared to adding up individual effects.

The caveat here "for complex traits" means things like height, for which we know there are hundreds to thousands of common variants which matter. Some things aren't like that, e.g. simple recessive gene effects are interactions! And, with more data, this may change.

Yeah, you see a lot of penalized regression techniques applied in that field. With that said, identifying useful non-linearity might just require moar samples.

I don't think it is entirely genetic. There are other factors like development within the mother's body, levels of hormones / chemicals, and so on which could contribute to a different sexuality later on.

This doesn't mean you can change it, and messing with chemical levels in the hope of finding a configuration which leads to "normality" would be very unethical in my eyes.

This is the ingredient most of this discussion is missing. People here seem to mostly believe "If it's not genetic then it's nurture", but that's not the case. It can be all sorts of other things.

My (rusty) recollection of the literature is that our best guess for what most causes homosexuality in human males is the hormone profile given in utero. This is not caused by the fetus's genes, but it will interact with the fetus's genes. Hormone levels could be influenced by the mother's genes[0]

I think what HN-ers are really interested in is, "How much control can we humans practically exercise over this?", the answer being "very little" in both the case of "genes control sexual preference 100%" and "in utero hormone levels control sexuality 100%".

0: https://en.wikipedia.org/wiki/Maternal_effect

Given the pretty strong birth order effects in play when it comes to gay men[0], you could probably stop about a quarter of the gay male population from being born just by banning consecutive male births, assuming you were willing to enforce a comically evil level of control over the population.

[0] https://en.wikipedia.org/wiki/Fraternal_birth_order_and_male...

> Given the pretty strong birth order effects in play when it comes to gay men[0], you could probably stop about a quarter of the gay male population from being born just by banning consecutive male births, assuming you were willing to enforce a comically evil level of control over the population.

A one-child policy with exceptions available if the first-born is a girl is a “comically evil level of control” that is hard to imagine being applied in the real world?

A one-child policy is ethically achievable by making the parents rich or otherwise making retirement care a non-concern. If you look at studies, rich people or people who have retirement taken care of are more likely to produce less children. A socialised retirement plan along with a good safety net for those unemployed would likely drive down population massively within a few generations.

I'm not a huge fan of China's one-child policy, no, and I'd be even less of a fan if they'd done it with the intent to eradicate gay men.

> There are other factors like development within the mother's body, levels of hormones / chemicals, and so on which could contribute to a different sexuality later on.

Those could still be influences if it was purely genetic. Biological sex is approximately fully genetic (not quite, but very close to it), but womb environment (I remember pH differences specifically being identified as an influence) play a role in which sperm, with which sex chromosomes, are most likely to survive and reach the ova.

https://en.wikipedia.org/wiki/Crocodile#Reproduction

On many complex species, the genere isn't defined by the genes.

I meant “in humans”; yes, there are places where that isn’t the case.

This sort of approach is known as Genome Wide Association study, finding the correlation of hundreds or thousands of genes that will predict a trait: https://en.wikipedia.org/wiki/Genome-wide_association_study

If you are interested in the topic, I recommend https://en.wikipedia.org/wiki/Blueprint_(book)

Gay-WAS

Interesting. The direct correlations seem to be weak, though.

I wonder if biology will turn out to have something comparable to the "hidden units" of machine learning, where less specific inputs from DNA combine in some way to produce intermediate - somethings? - that key specific functions. Probably something that will be explored over the next few decades.

The best you could probably do would be to match twins, so about 65%[1].

[1]: https://pubmed.ncbi.nlm.nih.gov/8494487/

This is an interesting idea. I wonder if we could then use something like [0] to trace back "gay-ness" as well as other "personality features" (I'm not sure what the correct term is here) to the exact genes.

One potential issue I see is that the DNA of any person has an arbitrary length, which poses some challenges in the design of a neural network. Traditionally, this has been solved with LSTMs or RNNs, but as far I know these are designed for data with a temporal dimension (such as text or speech, which progress with time). I'm not sure if that's true for DNA.

[0] https://arxiv.org/abs/1509.06321

Or.. there is no such thing as a gayness gene. Genetics are generally not a strong predictor for human behavior. Gestational environment, what sort of environment you grew up in, whether or not you've suffered any head injuries or have another developmental disorder, internal hormonal environment, the culture you live in... these are all much better predictors.

There are enough twin studies out there to rule out environmental factors as the sole cause.

Twin studies where the twins don't share gestational environment are...rare, to say the least.

You could of course argue that environment in the womb is largely genetic (and has a 50% correlation with the kid's genes, which is damn influential).

But once we open that door we're going down an extended phenotype rabbithole where you can start to talk about social interactions amongst non-immigrants being partially genetic because of correlations within regions, etc, and hilarity (with a healthy dash of idiocy) ensues.

The greater point being that saying something is either genetic or due to environment is always a pointless exercise because there is no separation, there are merely genes that spread without ever deciding how to allocate blame or credit for their successes and failures.

Which makes them interesting when one twin is gay, and the other is not; or when comparing outcomes for monozygotic twins vs fraternal twins.

Reply on news.ycombinator.com