Can you tell us about your background?
I currently work at the economics department of the VU. My research focuses on the genetic architecture of social science outcomes, which is a very broad category. I’ve mostly been working on the genetics of educational attainment (i.e. years of education), but I’ve also conducted Genome-Wide Association Studies (GWAS) of wellbeing, neuroticism, and depression. Methodologically, I mostly focus on polygenic indices.

GWAS of educational attainment is considered to be particularly affected by confounding. Which confounds are these?
Educational attainment is indeed a poster child for any confound that you can think of. The most important ones being population stratification, genetic nurture, as well as within and cross-trait assortative mating. There is also ascertainment bias, because educational attainment is correlated with a wide range of traits, including participation in genetic studies.

Is it fair to single out educational attainment here? Are these biases not also problematic for other traits?
It’s a fact that these biases affect educational attainment more than other traits. For example, educational attainment and genetics jointly stratify our societies, down to the scale of neighborhoods, making it difficult to disentangle true genetic effects from population stratification. This is probably less of a problem for, say, blood pressure. But, having said that, these same biases affect many other traits, especially behavioral and psychiatric ones.

Despite all these difficulties, you are still interested in the genetics of educational attainment? Why?
Not despite, but precisely because genetic studies of educational attainment are so complicated do I want to conduct them. We can use educational attainment to understand all these confounders better and then develop methods to account for them in any trait. But my main reason for being interested in it is that it’s correlated with all kinds of other traits that we deem very important, such as health outcomes and longevity. As such, studying educational attainment can teach us something about them too.

You were a coauthor of the educational attainment GWAS that was cited in the manifesto of the Buffalo shooter (https://www.statnews.com/2022/05/23/buffalo-shooting-ignites-debate-genetics-researchers-in-white-supremacist-ideology/). Did this affect how you think about your research?
The Buffalo shooting was clearly a very sad event. It did make me think more about the consequences of doing this type of work, but I don’t believe that we should stop this line of research. The problem is not that there are genetic differences between people and that these affect outcomes such as educational attainment. The problem is that these differences become a source for hate, which is not a scientific problem, but a societal one. There were white supremacists before this type of genetic research was possible; they will not go away if we were to simply stop the research. However, that does not mean that we shouldn’t be extra careful in how we communicate our results. We do things like creating FAQs and regularly talk to the media about how the results can and cannot be interpreted. This is important for the general public, but I believe it is unlikely to change the mind of a white supremacist.

You are the chair of the Polygenic Index repository – can you tell us about this repository?
As the name suggests, it is a repository for Polygenic Indices (PGIs). We receive genotype data from participating cohorts and for any given trait we perform a meta-analysis of GWAS summary statistics from multiple sources, including novel GWAS we conduct in UK Biobank or receive from 23andMe. Then we use this large meta-analysis to make PGIs for individuals in each cohort. These PGIs are sent back to the cohorts and made available to researchers with access to these cohorts. The first release of the Polygenic Index repository had 47 traits, but I am currently working on a second release with around 20 more traits. We have three main reasons for creating this repository. First, we can distribute PGIs based on the full 23andme data. Normally, summary statistics that include 23andme are not publicly available, meaning that only suboptimal PGIs can be computed. Second, it standardizes the computation of PGIs, which allows for a better comparison of effect sizes between PGIs, especially with the measurement error correction tool that we also released. Third, PGIs can now be used by researchers who themselves are unfamiliar with the software to compute them.

Some critics argue that focusing on genetics detracts from addressing systemic socio-economic and environmental factors that impact educational attainment. How do you respond to this perspective?
Until recently, only environmental effects on social outcomes were studied, because we didn’t have the data or tools to study the genetics. Now that we are able to study the effects of genetics, it suddenly distracts from studying the environment? That argument doesn’t make sense to me, especially since it’s really only a small subset of all of the social sciences looking at the genetics side of it.

What discovery or advance in genetics are you most excited to see in your lifetime?

It would be really exciting to see all of the Alzheimer’s genetics finally figured out. Alzheimer’s affects quality of life in a major way for a large proportion of the population. A better understanding of this disease would be really great.