Genetic structure correlates with ethnolinguistic diversity in eastern and Southern Africa

Abstract

Humans are believed to have originated in Africa, resulting in more genetic variation on the African continent 10 than anywhere else in the world; the average African genome has nearly a million more genetic variants than 11 the average non-African genome(Consortium and The 1000 Genomes Project Consortium, 2012). Africa is 12 also immensely culturally and ethno-linguistically diverse; while the rest of the world averages 3.2 to 4.7 ethnic 13 groups per country, African countries have an average of greater than 8 each and account in total for 43% of 14 the world’s ethnic groups(Fearon, 2003). Despite this diversity, African ancestry individuals are sorely 15 underrepresented in genomic studies, making up only about 2% of GWAS participants(Popejoy and Fullerton, 16 2016; Sirugo et al., 2019). Furthermore, the vast majority of African ancestry populations currently represented 17 in genetic studies are African Americans or Afro-Caribbeans (72-93% in the GWAS catalog and ≥ 90% in 18 gnomAD) with primarily West African ancestral origins(Martin et al., 2018). These resources thus currently 19 leave out the substantial diversity from regions of Africa that would be informative for human genetics. 20 21 Populations underrepresented in genetic studies contribute disproportionately to our understanding of 22 biomedical phenotypes relative to European ancestry populations. Despite their paltry representation in 23 GWAS, African ancestry populations contribute 7% of genome-wide significant associations(Martin et al., 2018; 24 Morales et al., 2018). African population genetic studies are especially informative given their unique 25 evolutionary history, high level of genetic variation, and rapid linkage disequilibrium decay(Tishkoff and Verrelli, 26 2003). This Eurocentric bias in current genomics studies and resources also makes African descent individualess likely to benefit from key genomic findings that do not translate fully across populations, contributing to 2 health disparities(Martin et al., 2019). In this study, we better characterize the immense genetic and 3 ethnolinguistic diversity in four countries in eastern and southern Africa, offering insights into their population 4 history and structure. Data are from 900 genotype samples that are part of the Neuropsychiatric Genetics of 5 African Populations-Psychosis study (NeuroGAP-Psychosis), a major research and capacity building initiative 6 in Ethiopia, Kenya, South Africa, and Uganda(van der Merwe et al., 2018; Stevenson et al., 2019) 7 8 Genetic variation in Africa has been previously described as following not only isolation-by-distance 9 expectations, but as being influenced by multiple interconnected ecological, historical, environmental, cultural, 10 and linguistic factors (Baker et al. 2017; Uren et al. 2016; Henn et al. 2012; Henn et al. 2016; Sikora et al. 11 2011; Creanza et al. 2015; Kolodny et al. 2016; Creanza and Feldman 2016). These factors capture distinct 12 variation from that tagged by genetics and can be informative for understanding population substructure. Better 13 characterization of the ethnolinguistic composition of these samples is a key initial step towards running well- 14 calibrated statistical genomics analyses including association studies. If ethnolinguistic variation tags additional 15 structure than that captured by geography, explicit incorporation of relevant cultural information into such 16 analyses tests may be the optimal strategy. In addition to the covariation of culture and genetics(de Filippo et 17 al. 2011; Karafet et al. 2016; Barbieri et al. 2013), individuals’ cultural environments influence how phenotypes 18 are expressed and whether assortative pairing impacts the distribution of traits(Coelho et al. 2009; Uchiyama 19 et al. 2021; Creanza et al. 2017). We measure how ethnolinguistic culture has changed in parallel to and 20 independently of genetics, which provides a foundation for the study of phenotypes of medical interest. In this 21 study, we explore the genetics of the NeuroGAP-Psychosis dataset, which comprises five collection sites 22 across four countries in Africa, and how individuals’ cultural affiliations and languages are related to genetic 23 variation across the continent. We also explore ongoing linguistic changes and consider the impact they will 24 have on the genetics of Africa