How wide and how tall? Genome Wide Association Studies in debate, from height to educational attainment and back – Ulrika Björkstén (Clare Hall; Sveriges Radio)

May 9, 2019 @ 1:00 pm – 2:00 pm
Seminar Room 2
Department of History and Philosophy of Science
Richard Staley

In April 2018 a study was published that claimed to have found evolutionary differences between human populations on different continents, with a larger selection in some populations for genes which could be linked to educational attainment (Racimo et al, _Genetics_, 208, 1565–1584). A few months later the same research group presented new results calling to question the very foundations of their own methodology (Jeremy Berg et al, BioRxiv, 23 June 2018).

In this paper based on a chapter in progress I trace how claims about evolutionary differences between human populations in the selection of complex traits were constructed from 2010 to 2018, starting with the publication of GWAS-data linked to height by the so-called GIANT-consortium in 2010 (_Nature_, 467, 832–838). This article was considered a breakthrough in GWA-studies of complex human traits, finding hundreds of genetic markers correlating to height differences between individuals. Although the authors carefully noted that these genetic markers could not be used for predictive purposes, but should rather be regarded as indicators of genetic loci suitable for further exploration of functional genetic links to height, this set of markers was subsequently used in a study investigating evolutionary explanations to height differences between different European populations (_Nature Genetics_, 44, 1015–1019). The article published in 2012 was hence one of the first to claim to have found active selection for a complex human trait which differed between populations.

After these two initial articles were published, several years of similar research followed on ever larger datasets, as the GIANT consortium grew. With the increasing power of new datasets, including genetic data from hundreds of thousands of individuals, the number of genetic markers that could be statistically linked to height differences between individuals also grew. These results were in their turn used as starting points for new studies of the evolutionary background to differences in height between different populations, notably between southern and northern Europeans. Simultaneously, the success of the 2010-study had sparked a whole new field of research applying GWAS to an increasing number of human traits, including one as complex and highly culturally dependent as ‘educational attainment’. In 2018 the field hence culminated in a study claiming to have found an evolutionary background to differences between human populations in the frequency of genetic markers that could be linked to educational attainment. The political implications of this claim sparked a controversy within the population genetics community (Novembre et al, _Genetics_, 208, 1351–1355) which led to a reconsideration of earlier results. Comparing the GWAS-results from one data set (GIANT) to another (UK Biobank) researchers found that the statistical link between certain genetic markers and height all but disappeared. The explanation given was that the GIANT-dataset suffered from unknown population stratification. This finding brought to question the methodology as such, since it uncovered a hitherto underestimated sensitivity to confounding factors.

Interestingly, the seminal article of the field, published in 2010, included a clear warning of using GWAS-results for prediction. Instead it stated that the genetic markers found should only be used as a starting point for further investigations of genetic functionality. This story hence unfolds as a case study of how scientific results take on new meanings as they leave their original setting and are interpreted by other researchers and implemented as starting points for new studies.

Another aspect of this case study is that the original datasets used had a very heavy bias of northern European genetic data. Recent studies have shown that genetic markers found by GWAS in one population cannot necessarily be transferred in a meaningful way to another population. This story hence also serves as a very concrete illustration of how a Eurocentric approach may skew scientific results.