Thread by @ewanbirney

Ethnicity, Ancestry Groups and Biology in humans; some thoughts triggered by @molly_przew threads and the recent Oxford paper on the likely mechanism of action for the COVID risk locus on chr3. TL;DR This is area is complex; racism +discrimination are real; biology is universal.

It's useful to remind yourself what some of these terms mean (or might mean). Ethnicity (also "Race" in US context) is usually defined via self identification - a person is given a number of options to tick, sometimes with hierarchy, and they tick one (or more) option.

These boxes are very culturally and nation-state defined. eg: The US has it's census terms (census.gov/topics/populat…); the UK has different terms (ethnicity-facts-figures.service.gov.uk/dashboards/eth…); Japan their own scheme; in France it is illegal to ask this question.

Ancestry is a complex word, with multiple meanings and usages; a good primer is here from @mathiesoniain and @aylwyn_scally journals.plos.org/plosgenetics/a…

I like this paragraph "...ancestry itself is rarely defined. We argue that this reflects widespread underlying confusion about what it means in different contexts and what genetic data can really tell us. This leads to miscommunication between researchers in different fields ..."

Ancestry groups (expanded to "continental ancestry groups") have come to mean a sort of scientific way to describe the broadest level of ethnicity, and rather dangerously confounded with this in many people's mind. I think this is deeply unhelpful.

To set the scene: humans exploded across and out of Africa only 100,000 years ago - we are a super young species - and like all species the fundamental relationship between us as individuals is a complex family tree, in our case stretching back into Africa.

Each part of our genome has it's own tree. So, for me, on one part of my genome (say top of chromosome 1) I have some 3rd cousins in Northern Ireland, some 5th cousins in Sweden, a 6th cousin in Mumbai, and a whole bunch of 12th cousins in Nigeria...

A different part of the genome, something different - say a 3rd cousin in Jerusalem, some 4th cousin in scotland, a 5th cousin in Morocco, and some 4th cousins in Mid-west America and some Congo 8th cousins.

So not only as we go back in time do we double the number of ancestors in our family tree (parents, grandparents, great grand parents); remember doubling will get you to a very big number very quickly (20 generations back, 1 million; 30 generations , 1 billion)

These billions of ancestors overlaps in complex ways (the same person being your great-great-great-great-XXX grandfather on one lineage but also you great-(x8) grandfather on another lineage) but also >>

the genome you have rapidly becomes sparse in this growing tree - many ancestors do not contribute to your genome. This exponential growing history, sparse sampling, different sparse sampling in different parts of genome means really rather quickly it just becomes a big tangle

Notice there are no groups here in ancestry and this tree. Groups don't help much in understanding genetics (they often get in the way) because the fundamental thing is a tree. Groups *are* important when we talk about culture including movement (as groups of people often move).

Although some humans don't move (and we can see that in their genome) some humans do move, often as groups, (and we can see that too in their genome) and over multiple movements the mixing of where genomes are vs geography gets doubly messed up.

There are many human stories here; some horrific, like the forced, awful transatlantic slave trade; some intriguing, like the fact that stonehenge looks like it was started by some people and finished by a different culture. One of my favourites is the blending in Europe.

Three pretty big migrations are the founding sources of much of the DNA of people currently living in Europe - of the original out of Africa hunter-gatherers; of the first technologists of the world - the farmers of Anatolia; and of a success Steppe based culture that came in.

These groups didn't just migrate into Europe; the out of Africa humans split out from humans that populated most of the non-African world; the farming technologists also went down the East African coast, contributing to groupings in South Africa; Steppe people also went east.

But this story of movement and mixing one can repeat at different times in human history and places, and although the ones we can see archeological and genetic traces of are the bigger (cultural) groupings, it happens in smaller groups down to the individuals who move.

But the mindset of "humans are a big tangled mess" goes against the societal gestalt feeling that "humans are different and these differences are meaningful" -

ie, I will tick the self identified ethnicity form presented to me in the UK with confidence (in my case "White British") and other people would gestalt recognise me as at least "White something" if they were asked which box I ticked. Surely we are confident of these groupings?

We are, but this is because we've placed one visible difference as a key grouping. If our boxes were hair colour we'd confidently tick a box (blond, brown, black, red) and other people would be confident about which box we'd tick. Would we have "mixed hair colour families"?

However, the gestalt ethnicity assessment, created and defined above all during the time of slavery and oppression by Europeans persists and is pervasive here. (I am not a historian here. I really recommend reading historical perspective here, eg >>

the nuanced book Black British A forgotten history by @DavidOlusoga amazon.co.uk/Black-British-…)

This societal, cultural overlay of family wealth, discrimination, structural racism / unconscious bias has *profound* effects on the biology of humans; just the straightforward access to healthcare and the trust between people and state (COVID throws this into stark light).

Armed now with this scene of both our genetics and our culture, an interesting question is what humans do we study when we want to understand humans - from mechanistic molecular biology through to how humans interact with healthcare systems?

Well, at a first glance, the simple answer is that we want to sample broadly and widely, and thus ensure we generalise well. However, logistics and practicalities often get in the way here, and one has to be a bit more nuanced, and here it is critical about the questions.

The more cellular and molecular you are the less one expects cultural components to impact the thing one is studying, though note there are still plenty of culture-induced environments (eg; teas from wild herbs that have the powerful mutagen Aristolochic acid)

There is a side note for genetics here; if you want to use naturally occuring genetic variation in your molecular/cellular dissection you are best off using samples with the highest diversity + most recombination - for humans, these are found in our birthplace, sub-saharan Africa

The more whole organism, societal the more care one has to take on ascertainment and generalisation. One *massive* issue is access to healthcare, from formal things (insurance) to cultural things (how easy do you find it to go to the GP) which is really complex on health.

As understanding how human biology impacts health is big part of what we want to do research on, this is really important. It is far better to be over-paranoid about this than not, and really have diverse samples to know things generalise.

So... full circle to the topics that triggered this long thread (and thanks for reading so far!). I agree with @molly_przew that we need to understand when and why we need to think about ethnicity / ancestry (I actually don't like the ancestry being a stand in for ethnicity)

Noting that human biology mechanistically is universal (though of course varies in interesting ways between individuals) and then human culture, including importantly racial discrimination, and the environments it sets up is very non-universal.

And to the COVID chr3 locus. This locus has a higher allele frequency in the Ganges delta - Bangledesh and Kolkata area. Although most of the people with ancestry from that area in the UK tick the box "South Asian" it is a much more focused subset and ultimately an individual

Although genetics like this shows that some of the variation in COVID outcomes is due to genetics, we can actually use it as powerfully to say the majority of the *disparity* in outcomes associated with ethnicity are not genetic

ie, it shows we're powered to see things, and they are not widespread (as expected by the messy tangle that is human genetics) to explain the BAME disparity in COVID outcomes; ie, the majority explanation is societal / cultural not genetic.

Thank you for reading to end!