Extended Data Fig. 1: The expected coalescence distribution based on the inferred demography matches the simulated input.
From: A likelihood-based framework for demographic inference from genealogical trees
(A) We simulated 100 equal-distant trees of 1000 haplotypes were simulated on a 30 Mb chromosome, under the same demography as in Fig. 2a. The demography is inferred by gLike on the true trees with default settings, and the expected coalescence distribution is computed by simulation of 10,000 trees under the inferred demography. The two distributions are highly consistent, except for small random fluctuations on the observed distribution. (B) The same experiment as in (A), but tsdate reconstruction is applied to the observed trees, the parameters are then inferred by gLike on the reconstructed trees, and tsdate reconstruction is again applied to the simulated trees under the inferred parameters. Vertical dash lines indicate t1, t2, and t3 in the simulated demography, corresponding to the time of the more recent admixture event, the more distant admixture event, and the split of three ancestral populations, respectively.
