File:F2 variant sharing across 1000 Genomes individuals.png

English: f2 variant sharing across 1000 Genomes individuals. Colours from white to blue to red show the number of f2 variants shared between each pair of individuals, normalised by the total in each row. Individuals are ordered by populations, but only by sample name within each population.

We applied our estimator to the phase 1 data release of the 1000 Genomes Project, which consists of whole-genome variant calls for 1,092 individuals drawn from 14 populations. We used two of the 1000 Genomes callsets; one made from sequence data, and one made using a dense genotyping array. Restricting our analysis to the autosomes, we extracted f2 variants from the sequence data callset, and then detected haplotype lengths around them (that is, the distances to incompatible homozygotes), using only the array data, to minimise the effect of genotyping errors. We then counted f1 variants on these haplotypes from the sequence data callset. From 4,066,530 f2 variants we detected 1,893,391 f2 haplotypes, with median genetic and physical lengths of 0.7 cM and 600 kb respectively. The median number of singletons spanned by each f2 haplotype was 3. Of the 1.9 million f2 haplotypes, 0.7 million were shared within populations and 1.5 million were shared within continents. Sharing of f2 variants largely reflects expected patterns of relatedness on a population level, and also reveals substructure in some populations, notably GBR.