an evenly spaced grid from 0 to 1) and infer the ancestry proportion α that minimizes F ST, as described previously [35],[36]. We compute the standard error of the ancestry estimate α via a bootstrap approach. We partition the set of SNPs into B disjoint blocks (e.g., B = 100), repeat the computation for SNPs in each block to obtain B different ancestry estimates, and compute the standard error as the standard deviation of these estimates divided by the square root of B. Standard errors of F ST estimates are computed in the same way. We note that the computation of F ST between two sampled populations is equivalent to the above formula for α = 0 or α = 1.
