Thread by @WiringTheBrain

Polygenic screening of embryos is here, but is it ethical? theguardian.com/science/2021/o… - great overview by @philipcball; my own thoughts follow...

In my view, polygenic scores are a statistical tool, not one with adequate precision in individuals to warrant use in embryo selection.

It is true that those in the very highest PGS deciles for some complex diseases have significantly higher risk than the population average. But any such embryos will also have sibling embryos with higher than average PGS...

...so any difference in risk between embryos is likely to be smaller than across the whole PGS range, and will remain probabilistic

Also, PGS for different complex diseases (heart disease, schizophrenia, diabetes) will be largely independent, meaning different sibling embryos will be slightly higher or lower for each

So, IMO, it's not quite snake oil, but it's not far off. Some would say that the predictions will inevitably improve, but they will - at best - hit the hard ceiling of heritability of these disorders

And in fact they're unlikely to get close to that as they only tag common variants, which only capture a fraction of the total heritability, which itself contributes only a fraction of the variance in who gets the disorder

As for selecting for traits like height or IQ or proxies like educational attainment, the same technical limitations apply, and you can add a host of additional ethical issues

You might say: let parents decide, where's the harm? Well, one additional problem is that we don't know where the harms might be - either biologically or sociologically

Having these polygenic scores to hand for different traits does not mean we understand the underlying biology - either of the whole set or of individual genetic variants.

We would be selecting based on one statistical outcome without clear knowledge of what it is we are really selecting on or what other things such variation might affect

The history of animal and plant breeding is rife with examples of unintended effects of prolonged selection - think of tomatoes selected for shelf life, but which lost all flavour in the process

So, the precautionary principle would seem to provide wise guidance on these issues.

As for the sociological issues, I don't see a real worry of a genetic super-class emerging among the rich - this kind of selection just won't work that well. Not well enough to justify doing IVF if you weren't already doing it

In fact, I would say it won't "work" for individuals at all, in the sense that the precision of the predictions is just too low. The only way it would "work" is the way animal breeders use it - in a breeding program, over generations

It's interesting that Steve Hsu, who generally positions this issue as one of individual parental choice, betrays his eugenic leanings when he says "The potential public health benefits are huge"

One gets the impression that he'd be glad to have this technology deployed at scale (in fact he says exactly that), such that the overall "genetic quality" of the population would be "improved"

And that gets into some much deeper ethical and moral issues about the relative worth and value of different types of persons - either in society, or as our own children

What's so interesting in this whole debate is that that question is a non-starter for many people but a perfectly valid one for others. I wonder if we could get a polygenic score for that?