We used MSMC5 to estimate changes in population size and split times. This analysis used two different phased genome datasets (using Shapeit v.265 and Eagle266). The details for the phasing are described in Supplementary Information 4. Chromosome 6 was excluded from the analysis owing to possible phasing errors in the HLA region. We used four haplotypes (two individual genomes) for estimating changes in population size in a population and eight haplotypes (two genomes from each of a pair of populations) for the estimation of population split times. We assumed a mutation rate of μ = 1.25 × 10−8 per site per generation and an average generation time of 29 years, as in previous studies8,19.
