For single-stranded library preparation, ancient DNA molecules are dephosphorylated and heat-denatured. A biotinylated adaptor oligonucleotide is ligated to 3′-ends of the molecules, which are immobilized on streptavidin-coated beads and copied by extension of a primer hybridized to the adaptor. One strand of a double-stranded adaptor is then ligated to the newly synthesized strand. Finally, the beads are destroyed by heat to release the library molecules (not shown).

Average sequence divergence and branch length differences between the Denisovan genome and 11 present-day humans represented as a tree. Divergence is reported as fraction of the branch leading from human to the common ancestor with chimpanzee, and in years, assuming a human-chimpanzee divergence of 6.5 million years ago.

Maximum likelihood tree relating the Denisovan genome and the genomes of eleven present-day humans, allowing one migration event (shown in yellow).

(A) Sharing of derived alleles among present-day humans, Denisovans and Neandertals. We compare all possible pairs of 11 present-day humans {H1, H2} in their “D-statistics”, which measure the rate at which they share derived alleles with Denisovans (x-axis) and Neandertals (y-axis). Each point reports ±1 standard error bars from a Block Jackknife. D-statistics are color-coded by geographic region. The D-statistic is not the same as the mixture proportion; it is also affected by quantities like the amount of shared genetic drift between the samples being compared. (B) Sharing of derived alleles that are absent in Africans among present-day humans, Denisovans and Neandertals. We enhance the power of the D-statistics by restricting to sites where 35 sub-Saharan African samples have the ancestral allele, and pooling modern humans by region to increase resolution (bars again give one standard error). Eastern non-African populations have significantly more archaic ancestry than European populations (Z=5.3 and Z=4.8 for the tests based on the Denisovan and Neandertal D-statistics, respectively).

(A) Heterozygosity shown by the distribution of the number of bases matching the human reference genome at sites sampled to 20-fold coverage. The Y-axis is scaled to show the peak representing heterozygous sites in the center. (B) Inference of population size change over time using variation in the time since the most recent common ancestors across the genome shows that Denisovans have had a small population size over the last few hundred thousand years compared with modern humans, but a similar demographic history earlier. The y-axis specifies a number proportional to the population size N_e. The x-axis specifies time in units of divergence per base pair (along the top in years, assuming rates of 0.5×10⁻⁹ to 1.0×10⁻⁹ per year). Thin red lines around the Denisovan curve represent 100 bootstraps, thus showing the uncertainty of the inference. (C) The small population size in Denisovans is reflected in a greater accumulation of non-synonymous sites (normalized by the number of synonymous sites), whether measured in terms of heterozygous sites in Denisovans vs. modern humans (ratio 2.0 –2.5), or the accumulation of divergent sites on the Denisovan lineage divided by modern human lineages (ratio 1.5–2.0). The analysis is restricted to non-synonymous sites predicted to have a possibly or probably damaging effect on protein structure or function.