The closer you live to someone, the more likely you are to share recent ancestors with that person. This is visible in beautiful geographic patterns that human DNA variation shows within continents, but also within countries.
I learned this during my PhD project in Amsterdam with professor Dorret Boomsma (2010-2014). During the data cleaning process, I got obsessed with trying to figure out what the larger patterns of genetic variation in our dataset meant. The key to answering that question turned out to be in the geographic map of the Netherlands. Simply plotting the ~4,500 Dutch individuals on their longitude and latitude coordinates and coloring them according to measures of DNA variation paved the way for several articles on the genetic structure of the Dutch population. We described how historical socio-cultural factors, such as religion and education, influenced migration and mating patterns that helped form, maintain, and rearrange the geographic distributions of Dutch ancestry differences in recent centuries.
In 2017, I traveled to Brisbane, Australia, to start working on a “short” side-project with professor Peter Visscher during a couple of free months I had in between jobs. It ended up taking two years between start and publication date, including what the editor called a “rather challenging peer review process”. As it should have been, since the claims we make have quite some implications on both a scientific and a societal level.
We started exploring the geographic patterns of genetic variation in Great Britain using the UK Biobank dataset. With DNA from ~450,000 people of European descent, this time we had the statistical power to take it a step further by looking beyond ancestry differences at genetic variation associated with heritable traits. Instead of working with just longitude and latitude coordinates, I took some online tutorials on how to use Geographic Information System (GIS) approaches, the “proper” way to work with geographic data. I then used that to study the geographic distributions of polygenic scores, which are measures that are predictive for specific heritable traits based on measured DNA.
We constructed polygenic scores for more than 30 heritable traits related to physical and mental health as well as normal psychological outcomes such as personality and educational attainment. After carefully controlling for ancestry differences, the polygenic score for educational attainment showed the highest levels of geographic clustering by far. The amount of geographic clustering of all other traits could be predicted by how many genes they shared with educational attainment. Importantly, the clustering of educational attainment alleles was increasig over time because people that left the poorer areas of the country, such as coal mining regions, had higher polygenic scores for educational attainment than the rest of the country on average.
One of the GIS tutorials included the Brexit Referendum results as a practice dataset. As an exercise, I thought let’s see if these regional polygenic scores are predictive for Brexit results? This was not meant to be part of the project initially, but to our surprise, it worked pretty well. Especially for the educational attainment polygenic score. Where did this genetic signal for Brexit come from? I decided to find out by running a genome-wide association study on the regional Brexit measure, which also worked surprisingly well. We repeated this for a bunch of other regional measures, including more election outcomes and publicly available measures for socio-economic and health outcomes. On a regional level, the regional health outcomes showed an unexpectedly weak signal for health-related genes, but very strong signals for educational attainment genes, which we figured are just a very good proxy for the big differences in living circumstances between the richer and poorer regions. The living circumstances of poor neighborhoods that are making its inhabitants less healthy are probably the same that are driving people with a higher education to migrate away and the people that stayed behind to vote against the political status quo.
With this paper, we ended up showing how social stratification in a modern meritocracy makes it increasingly less likely that someone close by shares recent ancestors with you and more likely that they share a genetic talent for socio-economic success.