Knowledge of the complex history of Asian populations informs optimal sampling for larger-scale biomedical sequencing efforts. We applied standard approaches for detecting recent positive selection, quantifying the population structure and inferring the history of the different populations, including principal component analysis18, multiple sequentially Markovian coalescent (MSMC)19, ADMIXTURE20, FST, uniparental analyses and the analysis of the Y chromosome and mitochondrial haplogroups (Fig. 2, Extended Data Fig. 1 and Supplementary Information 3–10). Our results generally recapitulate the broad inferences of previous studies, and ADMIXTURE plots show complex structure within south and southeast Asia (Fig. 2a). In particular, India, Malaysia and Indonesia contain multiple ancestral populations as well as multiple admixed groups. On the basis of MSMC cross-coalescence rates, which reflect the increase in coalescence times of haplotypes sampled from different populations relative to haplotypes sampled from the same population19, we estimate that the oldest population splits in southeast Asia and Oceania involve Melanesians and/or Negritos, who show a substructure from approximately 40 thousand years ago and evidence of separation around 20–30 thousand years ago (Extended Data Fig. 1b and Supplementary Information 3).
