Abstract
Many of the modern human populations inherited a fraction of their genomes from archaic humans such as Neanderthals and Denisovans following ancient introgressions. Non-African populations have around 1–3 \(\%\) of Neanderthal ancestry, while some Oceanians and south-east Asians have in total up to \(6\%\) of Neanderthal and Denisovan ancestry. Accurately inferring genome tracts introgressed from these archaic population is important for reconstructing the history of human past, studying natural selection of these archaic components which in turn might be important to understanding functional role of their genes. Here we present a new Hidden Markov Model based approach to accurately detect these archaic tracts in modern humans by simultaneously using information from an unadmixed reference outgroup and archaic genomes. Using simulations we show that it outperforms a state of the art method both by precision and recall while being robust to the misspecification of population scenario parameters.
