We've got a new preprint out! Looking into the prediction of skin, eye and hair colour in a diverse cohort beyond just European ancestry, especially into Native American ancestry, really highlights the value of studying genetic diversity. https://twitter.com/biorxiv_genetic/status/1337163042333876226

Most large biobanks, and also most available pigmentation prediction tools, are Eurocentric. At least for eye and hair colour, they're not too off as Europe has the biggest variability for those traits. But skin colour is a different game!

It is has a huge diversity worldwide, and recent GWASes gave several interesting genes in Africans, East Asians and Native Americans. True to this, we show that training your model on Native Am individuals is really important for prediction in Latin or Native Americans.

In our GWAS with @JavierMenRev we showed a new skin colour hit in the gene MFSD12, seen only in East Asians and Native Americans, was possibly under selection, and showed a correlation between solar radiation and its allele frequency worldwide. Here we take that story further...

by linking solar radiation with predicted skin colour. We didn't have an independent geno-pheno Native Am dataset to validate our model, but we show that our predicted skin colour in highly Native individuals correlates much better with solar radiation across the Americas.

Lastly, we show that even training the same model on a more diverse cohort brings improvement. So, as many have said, expanding our research to more diverse populations can really enrich our knowledge. Of course, has to be done ethically & morally, as @kstsosie and others stress.