ghost retweetledi

No. I think people deserve to know how these things work before you poh on them. I'm sure you know much of this, but the general public does not.
There are hundreds of so called "rare variants" of which if you develop something through idiosyncratic mutation ("de novo" as opposed to "germline") which drop your IQ by a substantial amount. This explains almost all of the variance between fraternal siblings.
Prior PGS before deep sequencing only identifies common variants, which do not contain useful information, because in general fraternal siblings do not have a lot of variability in their common variant background, which represent their common *ancestral* genetic lineage, and even if they do it's minimally informative on an individual basis. However, things changed in the last 5 years where suddenly a huge amount of deep exome sequencing data emerged, and hundreds of new rare variants are identified, such that the so called "missing variance" that is estimated in twin studies is no longer missing if you tabulated all the rare variants. An intuitive way to think about this is that rare variants are like harmful switches that drop your IQ. And the screens find them if they exist, and hence optimizes offspring IQ within a cohort of possible siblings. The effect of this depends on the common variant background but can be quite substantial.
The real controversy is how MUCH that effect is related to the present ancestral background. I.e. if your common variants say you are gonna be IQ 100, even if you don't have any of the harmful rare variants, maybe it won't make much of a difference if you detected a couple. And there are yet not understood nonlinear interactions between common and rare variants, and people are working on this with ML. Also many of us have these rare variants and perhaps have lower IQ than we are "supposed" to have, another poorly understood finding. Nonetheless, the new results are genuinely new and impactful.
This is also the same architecture, for highly heritable but "polygenic" psychiatric traits like schizophrenia and autism, and explains much of the "missing variance" in the prior PGS studies. But there seems to be less controversy surrounding the same techniques optimizing for those traits.
English





















