Yet again Ethnicity != Genetics. (evergreen tweet). Despite the fact that skin colour is predominantly genetic, ethnicity classification, which often uses skin colour as part of the gestalt assessment of ethnicity is *not* a proxy for overall genetics

There are many ways to think about this. One simple one is that skin colour genetics is not a million miles away from hair colour genetics (its a bit more complex, but not so much) and we don't think of hair colour as somehow definitive of the rest of the genome

(nor do we routinely socially hassle people who have red hair about "which part of Ireland are they from originally" or "what is your heritage" because... we culturally get that hair colour is just hair colour. We nearly always see the individual first, not the hair colour first)

Another way to think about this is that ethnicity classifications (themselves hugely different country to country) is a process of ticking boxes (sometimes writing in "other" line, which is then often not used). It's a 1-dimensional, ~10 bits of information about someone

Genetics on the other hand is a formally ~3 billion dimensional thing, but occupies thankfully a far small space (let's call it ~300,000 dimensions, around haplotype blocks; it's a complex beast) and it doesn't squish down easily beyond that.

The dimensionality already shows the limits of this. One can sometimes reasonably go from genetics (~300,000 dimensions) => ethnicity (1 dimension, 10 bits). You can't go backwards. It just *doesn't work*.

(we can *force* a squish down into 2D PCA space or into some UMAP/tSNE 2D space, and this can sometimes vaguely correspond to ethnicity labels, but this squish is genetics => 2D => ethnicity labels and it is not reversible)

Another way to think about this is the common "Mixed race" ethnicity category in many systems. This sort of "its a complicated" is not some small grouping of people - it is *all* of the Hispanic/Latinx grouping, and *all* of Afro-Caribbean; it is just a perspective time

And this time perspective just goes back - Cape Coloureds, Goans... and then... Europeans as mixtures of founding people. The whole move, have children, move over the many generations have happened again and again by our species.

And you can also see it in our population genetics, human archeology and the formation of our species - from physical sites that we discover showing movements of people and cultures, to our language sharing, to the way to trace individual histories via genetics

These defy - make a mockery of - "ethnicity" of people as categories. It is just ... not how humans peopled the world; categories of humans just ... are a terrible way to think about human genetics and terrible way to think about humans overall.

...but... ethnicity classification is quite pervasive in our society, because of our behaviour - people's gestalt's assessment of other people and other people are consistent inside a society

So when we see differences on the basis of ethnicity between groups of people it is unlikely the main drivers of this is genetics - shifts in allele frequencies which are skewed between ethnicity *might* be part of explanation, and sampling a diverse set of people is important

much of that diversity is as important (probably more important) about the societal / social differences associated with ethnicity - some of this is economic, some other aspects of our society, for example education.

It is complex and messy - this is the well trodden realm of epidemiologists and sociologists - and genetics can probably illuminate some things, for example using weak genetic effects as natural experiments (my other evergreen love of mendelian randomisation).

but, don't think that shifts outcomes/behaviours/diseases between ethnicities is somehow likely to be driven by genetics, or be inherent. Just does not make sense.