Good news- Spectrum 10k study has been postponed

I only just learned about the study from this post, but... and I know this will be controversial, but isn't greater understanding of autism a good thing for us? I understand the fears about eugenics, but so far as I know there isn't any technology sophisticated enough to choose what genes to "turn off" before birth. The closest I'm aware of is something called CRISPR, but from documentaries I've seen on the subject it cannot be used to make any kind of genetic selections about something like autism, or any form of learning disability or personality trait.

I'm happy to be corrected by anyone more educated than me on this, but as someone with a lot of issues associated with my diagnosis of autism spectrum I only feel disappointed. 

They won't turn off the genes before birth, they'll just terminate pregnancies that would result in autistic children, just like they do with tests for other disabilities.

Autism is known to be complex, Down's syndrome has a 'simple' cause - an entire extra chromosome. Down's is easy - relatively - to screen for, Autism, being, as we already know, multifactorial will be almost impossible to create a simple screen for. Plus we know that some of the genetic factors leading to autism are beneficial, so removing autistics from the community would be counter-productive. All that stymying a study conducted in a liberal demorcacy (or several) will result in, is something similar, but worse, being conducted in China or somewhere else where ethics are non-existant. I am far from happy at this news.

Well the paper I quoted was actually a meta analysis of several studies but either way if autism is at least 50% heritable (and its probably a lot more) there is every reason to think many cases could be inferred from genetic analysis alone. Just because the same level of genetic analysis necessary to infer autism could possibly infer intelligence it doesn't mean parents would be offered that complete spectrum of information. As you say people might feel giving parents predictions of intelligence, even crude ones, might be a bit eugenics like but sadly far more people would be comfortable passing on a prediction of autism. 

At the end of the day any autism 'test' would be a product with a one or 2 number output generated by a computer program no one outside of the manufacture really understood, certainly not the doctors giving test results to parents.

They might use something like ensemble machine learning on the gene SNP results. X% of the neural nets say autism, Y% are indeterminate, this is associated with Z% likelihood of autism. Something like that is what they'd get back. Sure you could get the same SNP results and run them through a different set of appropriately trained neural nets to get a prediction of intelligence ... but even if anyone creates such a tool it wouldn't be offered clinically.

The point about the paper on micro array screening I quoted is they were arguing for its adoption to replace screening that is already done using simpler methods. Let me put it this way. The chromosome micro array analysis is now common enough hospitals have information pages for parents explain why their children are being given the test. (www.nationwidechildrens.org/.../microarray-analysis-test) Biotech is getting cheaper. In ten years expect micro arrays to be routinely used in health care.

That is one study, in statistics -dependent studies a single study is rarely definitive. This paper gives a rather different spin, though it found genetic factors to be of more importance than environmental: https://pubmed.ncbi.nlm.nih.gov/25738232/

We know that for a small number of people there are certain genetic deletions and duplications that cause autism. The problem for most autistics is separating the wheat from the chaff, deleterious from non-deleterious alelles and other genetic differences. That is going to be very difficult. As autistic traits are entirely embedded in humanity at large, I thinkthat a straightforward, "there is an X% chance your child will be autistic" algorithm is unlikely. More likely will be, "your child has an  X% likelihood he will be an excellent visual artist, a Y% chance he will be a gifted mathematician and a Z% chance he might be autistic. How do you want to proceed?" This is far too close to "designer baby" eugenics for it to be acceptable to most societies. The interrelationship of autistic and gifted traits will make extirpating one without extirpating the other almost impossible.

I think that microarray screening will be expensive, at least in the short to medium term, and this will restrict its diagnostic use. The paper you cite was not using microarrays in a normal clinical setting - it was not a service - and it was looking at quite obvious departures from normal foetal development. Autism is considerably more complex and subtle.

That’s not my understanding and tbh the research Ive read doesn't support it https://pubmed.ncbi.nlm.nih.gov/26709141/

not having a simple yes no answer isn’t an issue. If an algorithm can tell parents ‘based of your child’s genes there is an X% chance he’s autistic.’ Some parents will still abort on that basis.

as for micro arrays for prenatal screening this is already very much in the works for other conditions bmcpregnancychildbirth.biomedcentral.com/.../s12884-020-03368-y

That’s not my understanding and tbh the research Ive read doesn't support it https://pubmed.ncbi.nlm.nih.gov/26709141/

not having a simple yes no answer isn’t an issue. If an algorithm can tell parents ‘based of your child’s genes there is an X% chance he’s autistic.’ Some parents will still abort on that basis.

as for micro arrays for prenatal screening this is already very much in the works for other conditions bmcpregnancychildbirth.biomedcentral.com/.../s12884-020-03368-y

Well the paper I quoted was actually a meta analysis of several studies but either way if autism is at least 50% heritable (and its probably a lot more) there is every reason to think many cases could be inferred from genetic analysis alone. Just because the same level of genetic analysis necessary to infer autism could possibly infer intelligence it doesn't mean parents would be offered that complete spectrum of information. As you say people might feel giving parents predictions of intelligence, even crude ones, might be a bit eugenics like but sadly far more people would be comfortable passing on a prediction of autism. 

At the end of the day any autism 'test' would be a product with a one or 2 number output generated by a computer program no one outside of the manufacture really understood, certainly not the doctors giving test results to parents.

They might use something like ensemble machine learning on the gene SNP results. X% of the neural nets say autism, Y% are indeterminate, this is associated with Z% likelihood of autism. Something like that is what they'd get back. Sure you could get the same SNP results and run them through a different set of appropriately trained neural nets to get a prediction of intelligence ... but even if anyone creates such a tool it wouldn't be offered clinically.

The point about the paper on micro array screening I quoted is they were arguing for its adoption to replace screening that is already done using simpler methods. Let me put it this way. The chromosome micro array analysis is now common enough hospitals have information pages for parents explain why their children are being given the test. (www.nationwidechildrens.org/.../microarray-analysis-test) Biotech is getting cheaper. In ten years expect micro arrays to be routinely used in health care.

That is one study, in statistics -dependent studies a single study is rarely definitive. This paper gives a rather different spin, though it found genetic factors to be of more importance than environmental: https://pubmed.ncbi.nlm.nih.gov/25738232/

We know that for a small number of people there are certain genetic deletions and duplications that cause autism. The problem for most autistics is separating the wheat from the chaff, deleterious from non-deleterious alelles and other genetic differences. That is going to be very difficult. As autistic traits are entirely embedded in humanity at large, I thinkthat a straightforward, "there is an X% chance your child will be autistic" algorithm is unlikely. More likely will be, "your child has an  X% likelihood he will be an excellent visual artist, a Y% chance he will be a gifted mathematician and a Z% chance he might be autistic. How do you want to proceed?" This is far too close to "designer baby" eugenics for it to be acceptable to most societies. The interrelationship of autistic and gifted traits will make extirpating one without extirpating the other almost impossible.

I think that microarray screening will be expensive, at least in the short to medium term, and this will restrict its diagnostic use. The paper you cite was not using microarrays in a normal clinical setting - it was not a service - and it was looking at quite obvious departures from normal foetal development. Autism is considerably more complex and subtle.