Hacker News Reader

Comment by mlboss | Hacker News Reader

> Autism is genetic. You’re born autistic and you’re autistic for life. Most autistic people are autistic because of genes they got from their parents. This is why autism often runs in families. Much less commonly, other factors can play a role, but usually in combination with genes from parents. There isn’t a single ‘autism gene’. Thousands of genes combine to make a person autistic. Scientists have identified many of these genes, but they think there are still many more to find. And there is no genetic test for autism. We don't fully understand how genes combine to make a person autistic. It is complex and no simple test can give an accurate answer. Genetic testing cannot predict whether a child will be autistic or not, and it can’t be used to diagnose someone as autistic.

Autism is genetic but we cannot do genetic testing and prove that you have autism. Seems like a very circular reasoning.

View on HN

Genetics is complicated enough that both things can be true without circular reasoning.

I've studied genotype to phenotype mechanisms for a very long time (but from the lens of molecular biology, biochemistry, and biophysics, rather than genetics). The language of genetics- especially in the popular press- is extremely loose and ambiguous, and people trying to understand the genetics of autism using the popular press (or even 8th grade biology genetics) are going to be very confused.

Basically in medicine and biology in general it's best to be flexible- avoid using logical reasoning, instead adopt probabilistic mindset, and assume that the entities you're reasoning about are extremely complex and you only have a small amount of noisy, partial information that is in a highly encoded feature space. You can often measure relationships or statistical linkages that appear significant (even after multiple test correction) but rarely discern the true causal relationship between the feature vector (genome) and the outcome (autism).

One of the biggest surprises in modern genetics is that many phenotypes are determined by the sum of effects of many gene products, including many complex interactions between those gene products. Some phenotypes are linked to 1000+ genes, and there are "only" ~25K genes in the human genome. The whole thing is just crazy and our ability to test for complex phenotypes continues to be very limited.

Indeed. Thanks for the great explanation. Honestly, the complexity would be very annoying if it wasn’t also invigorating and fascinating.

The interconnections between genes has led to proposals of “omnigenic” models, the idea that all genes expressed in a cell affect every given trait of the cell, to some extent.

When it comes to something like autism, it’s a word covering a vast number of traits and behaviors of many different cells that gives rise to that set of effects as seen from outside.

And the issue with genetic testing is, we often only look at specific regions of the genome. Even when we do whole genome tests, and compare whole genomes of autistic folks, what we find is no single origin but many overlapping patterns. No definitive set is found because we are looking at complex processes that have many nodes where mutations can mildly tweak the function. Many such functionally relevant mutations can chain together to give you the signatures of autism, and no two are the same.

All these reasons also point to why we are increasingly realizing autism isn’t a “disease” with a “cure”. It’s a way in which people can be people. It’s time society accepted that and expands itself to accommodate.

It seems only natural for nature to have it like this. It’d be hard for evolution to optimize things if the genotype to phenotype mapping was sharp, with phase transitions and such

How do geneticists go about identifying alleles associated with autism? Seems like the effect size would be very small for most of them, and given the relatively small sample population of humans that can be practically sequenced for this, it seems difficult to separate true signal from statistical noise.

Also is it actually just “a sum of cumulative effects of these alleles crossing a threshold causes autism” or more complex like “Allele A and B together has a very significant effect, A and C has a moderate effect but A, B, and C together is has no effect”

What's an allele? There is no real answer to that question. Or rather, the concept of allele is highly abstract and does not correspond well to actual genomic sequences.

Predicting phenotypes is not a linear function of the genome. It's a highly nonlinear function, affected by complex generative processes, feedback control, homeostasis, and many other active control mechanisms. With that said, you could come up with a proxy score that was a function of the genome and had predictive value for autism (ideally better than binary, with a description of the predicted autistic traits somebody showed).

Biology is the only field I've worked in where both "necessary but not sufficient" and "sufficient but not necessary" can both be the case at the same time.

I understand that they found strong enough evidence that it's a genetic condition, but at the same time it's not one single gene, it's a combination of many that combine to create the full spectrum, hence no easy genetic test. They'll get there in time as technology and research evolve. I don't see any circular reasoning.

EDIT: they have a FAQ about it: https://www.autism.org.uk/advice-and-guidance/about-autism/w...

Height has a strong genetic component, but there's no test for height.

A meter?

There are ways to predict a child’s height as an adult that are very accurate.

Most countries track growth curves, and ask parents about what their height is to correlate data with. One can also use bone age assessment.

But can you do it only having their DNA?

Lemme try again: My LLM hallucinates paper citations. I have a strong suspicion that this results from the parameter regime. Is there a test for paper citation hallucination?

I think we agree and you replied to the wrong comment here.

i.e. You can look at family history and determine that autism runs in families, and by controlling for environmental factors you can conclude that it's very likely to be genetic, yet you still don't have a genetic test for it.

Which is very different from measuring parents' heights and predicting the height of their child. Similarly you could look at 2 autistic adults and conclude that their child would almost certainly be autistic too.

The comment you're replying to isn't talking about predictions now, isnt it?

If your C code causes a segfault, is it fair to reason that there's a memory management error in the code? Does that make it practical to find where, so that you can just add a check for it?

What if it only happens on a single machine?

Memory defaults can also cause segfaults.

what if segfaults only occurred when the errant path was executed in manners not initially designed for? or alternately, if that code path never got touched, would we ever have known about the segfault? should we keep focusing on segfaults or look at the ways that programs are being operated out of scope?

There are two fallacies in your analogy.

Segfaults are precise symptoms - they occur only when there's a memory issue. Autism isn't precise, is a collection of characteristics that manifest differently in each person.

Genetic is, for the purpose of this conversation, a precise diagnosis - means you carry parent genes that will cause the issue. Memory management isn't: what caused it? Code error? OS error? Processor? Is it a bad code path or a unexpected exrtenal state?

Finally, an analogy of your analogy: you measure someone temperature and it's 40C. You diagnose a fever.

Segfaults are fever, just a symptom.

Are we still talking about testing for autism?

I should have known better than to try an analogy here

The analogy is "just because most X are caused by Y, doesn't mean you can easily test for X by looking at Y"

I don't think your first point (how specific/precise are the symptoms of X) changes things, and I think your second point reinforces the analogy rather than weakens it. Most segfaults are coding errors and not OS or hardware errors ("look for horses, not zebras" / "SELECT isn't broken"). Even good medical tests have false positives and false negatives

> Most autistic people are autistic because of genes they got from their parents. This is why autism often runs in families. Much less commonly, other factors can play a role, but usually in combination with genes from parents.

That seems like a pointedly naive viewpoint. You can have clearly predictive population level statistics about prevalence in family groups without clearly understanding the (very messy and unbelievably complicated) way that genetics shapes brain development.

Some of what you've quoted highlights my own belief on some of this:

Autism is also multiple things.

That may provide some insight about how there is not a "single `autism gene'" and how they hedged a bunch of these statements with "most" or "usually".

Maybe some day when we as a species understand brains better, some people will get a more precise label than "on the autism spectrum".

> Autism is also multiple things.

This rings true to me. As someone with an autistic loved one, I obviously think about it a lot.

If it were as simple as there being some simple genetic basis that gives you a single thing called autism, I would expect it, honestly, to have naturally died out long ago, just because it makes so many things harder for people that it cannot possibly have not had a disadvantage to reproduction.

I would expect it's not caused by genes themselves, but by control DNA. That's where a lot of the subtle structural stuff happens. But that's also hard to test for. It's like how you can't test how tall someone is going to be.

I've also read ages ago in Scientific American that the difference between our brains and chimp brains is almost entirely in control DNA, not in genes. It matters a lot, but is often overlooked. (It used to be called "junk DNA", before people realised how important it was.)

My understanding is that having certain genes makes conditions like Autism more probabilistic than deterministic.

can't wait until large scale datasets and LLM powered studies either debunk or confirm and quantify the effects objectively.

not sure why the downvotes?

I predict it would be possible to quantify the effects objectively, as a thought experiment: assuming functional proteins, gene regulation is about binding affinities of cellullar chemical species to promoter regions, so 200k promoter regions x number of chemical species present in humans give us the number of binding affinities that describe the behavior of human cells (essentially all cell types excluding some germ-line cells, but including brain cells).

the thought experiment: apart from ethics nothing prevents humanity from going to all the remaining wild-type or near wild-type humans (when did 5% of the local population start wearing glasses? as a measure of wild-type-ness), and collecting as many individual genomes as possible, for a reference data set for the table of binding affinities. For each individual genome there is 1 table of binding affinities. Now we start averaging the value of each location in the table by averaging this table over all the collected genomes.

So theoretically we could collect the bulk of the gene regulatory information / skills encoded in the genome for wild-type humans. Next we can traverse the medical literature on autism and collect all the reported or suggested features, survey answers, diagnostic indicators for the full autism spectrum, and for each population in such a separating indicator, fetch example genomes from the purported indicated and unindicated groups, and compare their averaged binding affinity tables until you resolve or exclude any genetic correlation.

Or people downvote because I can't wait for science to ... progress?

Opened this link to quote that. The level of reasoning is "trust me bro". Even if some things must be simplified for uneducated reader, the quote from above should be embarassing for author to write

Reply on news.ycombinator.com