Extended Data Fig. 2: Twigstats optimal cutoff.
From: High-resolution genomic history of early medieval Europe
a, Theoretically computed z-score of f4(PO,P1,PX,P2) at a single genomic locus (Supplementary Note), assuming PX is admixted between P1 and P2 at time 0.004 (in units of 2Ne generations), e.g. corresponding to 100 generations with 2Ne of 25,000. Sources split at time 0.02. b, The theoretical fold-improvement of the best Twigstats z-score of f4(PO,P1,PX,P2) relative to the z-score obtained with regular f4-statistics. We use the same parameters as in a, but vary source split times to illustrate the improved power for mixtures involving more closely related groups. c, The optimal Twigstats cutoff time as a function of the source split time and the ratio between the optimal cutoff time and source split time. d, Comparison of z-scores computed using Twigstats to the corresponding theoretical values shown in a.
