A continuum of admixture in the Western Hemisphere revealed by the African Diaspora genome

Abstract

The African Diaspora in the Western Hemisphere represents one of the largest forced migrations in history and had a profound impact on genetic diversity in modern populations. To date, the fine-scale population structure of descendants of the African Diaspora remains largely uncharacterized. Here we present genetic variation from deeply sequenced genomes of 642 individuals from North and South American, Caribbean and West African populations, substantially increasing the lexicon of human genomic variation and suggesting much variation remains to be discovered in African-admixed populations in the Americas. We summarize genetic variation in these populations, quantifying the postcolonial sex-biased European gene flow across multiple regions. Moreover, we refine estimates on the burden of deleterious variants carried across populations and how this varies with African ancestry. Our data are an important resource for empowering disease mapping studies in African-admixed individuals and will facilitate gene discovery for diseases disproportionately affecting individuals of African ancestry.

Figure 1. Whole-genome sequences of African-admixed populations in the Americas.

(a) Geographical location of 16 CAAPA sites and estimates of global ancestry across 642 samples from North, Central and South America and Africa. The transatlantic slave trade is illustrated for each colonial power, along with beginning and end years of the transatlantic slave trade for British/North American, British, French and Spanish Caribbean, and Portuguese/South America. The date of abolition of slavery noted for each country participating in the transatlantic slave trade. The bars depict the relative proportions of African (blue), European (red) and Native American (green) contribution at each CAAPA site. (b) Percentage of SNVs within MAF categories (1=singletons, 2=MAF<1%, 3=1%≤MAF≤5%, 4=MAF>5%) across all CAAPA sites illustrating discovery of novel variants (that is, those not previously annotated in dbSNP) across all ranges of MAF. (c) Site-frequency spectrum of known and novel SNVs within CAAPA. (d) De-convolution of novel alleles by ancestral background in CAAPA using a paired t-test illustrates an excess of novel alleles occurs on the African/African background in contrast to the European/European and Native American/Native American background.

Figure 2. Genomic portraits of admixture heterogeneity within and between populations.

(a) Estimates of global ancestry of the 642 individuals using ADMIXTURE analysis on a set of 113,090 LD-pruned SNPs and 3 ancestral reference populations (CEU samples Utah from TGP to represent European ancestry; YRI Yoruban samples from TGP to represent African ancestry; and Native American samples from Mao et al.46). (b) Principal component analysis (EIGENSOFT48) using this same set of SNPs and ancestral reference populations illustrating the two main axes of genetic variation in all 642 samples. (c) Heat map of doubleton sharing by population; colour is based on the percentiles of the number of doubletons per individual-pair from the same population or from different populations. (d) Correlation between autosomal and X chromosome admixture estimates with the identity line in grey (population membership defined as in b).

Figure 3. High levels of identity by descent indicate a bottleneck unique to the Honduran Garifuna population.

(a) Density plot demonstrating elevated pairwise IBD across the Garifuna sample summed across the autosomes. Note: distribution filtered to remove first degree relatives (b) Skyline plot of effective population size through time in the Garifuna, as measured from pairwise IBD using the program IBDNe. Line represents maximum likelihood inference, with shaded region the 95% confidence interval determined via bootstrap.

Figure 4. Admixture dynamics influence characteristics of deleterious variation defined by PhyloP_NH.

(a) Correlation between the number of total derived alleles, heterozygotes and derived homozygotes of deleterious sites and African ancestry for all samples within CAAPA. (b) Correlation between the number of deleterious derived alleles and African ancestry for all samples within CAAPA by coding and non-coding sites. (c) Distribution of PhyloP_NH scores for coding, and non-coding deleterious sites.