Extended Data Fig. 8: Replication previous Viking Age ancestry modelling.

a, P-values of 1-source qpAdm models with target groups shown as rows and source groups shown as columns, replicating Extended Data Fig. 5a of ref. ⁶ Left column uses p-values obtained from ref. ⁶. Middle and right column correspond to newly computed p-values in a qpAdm using, respectively, all SNPs and Twigstats-2000. Outgroups are YRI, CHB, DevilsCave_N.SG, WHG, EHG, Anatolia_N, Yamnaya, Estonia_CordedWare.SG (Supplementary Table 1). We excluded Denmark_IA.SG and England_Roman.SG from the rotational scheme as these groups overlap in ancestry with England_IA.SG and Norway_IA, respectively. Only samples with coverage exceeding 0.5 are used. For each target group, the source group with the largest p-value is shown with a black circle. b, qpAdm models of ref. ⁷ where modern populations are used as sources. As in ref. ⁷, we show ancestry proportions averaged over individuals in each group, where for each individual the model with the smallest number of sources and largest p-value is chosen. c, Replication using the same target samples as in b. We fit a maximum of two sources and choose the model with the smallest number of sources and largest p-value, requiring p > 0.01 for 1 source and p > 0.001 for 2 source models. The set of individuals used in b and c are identical and are comprised of targets with an accepted model in all SNPs and Twigstats-1000, removing 15 of 167 individuals. We additionally remove 17 individuals that did not have a feasible model in ref. ⁷.