This PDF file includes:

• Supplementary Materials and Methods
  
• fig. S1. Calibrated age range of the two human specimens from Devil’s Gate (OxCal version 4.2.4).
  
• fig. S2. Damage patterns for samples from Devil’s Gate.
  
• fig. S3. Sequence length distribution for samples from Devil’s Gate.
  
• fig. S4. Outgroup ƒ₃ statistics on PMDtools-filtered data.
  
• fig. S5. PCA on all SNPs using the worldwide panel.
  
• fig. S6. PCA on transversion SNPs using the worldwide panel.
  
• fig. S7. PCA on all SNPs using the regional panel.
  
• fig. S8. PCA on transversion SNPs using the regional panel.
  
• fig. S9. ADMIXTURE analysis cross-validation (CV) error as a function of the number of clusters (K) for the regional panel using all SNPs (top row) or transversions only (bottom row) and with (left column) or without (right column) MapDamage treatment.
  
• fig. S10. ADMIXTURE analysis CV error as a function of the number of clusters (K) for the world panel using all SNPs (top row) or transversions only (bottom row) and with (left column) or without (right column) MapDamage treatment.
  
• fig. S11. Outgroup ƒ₃ scores of the form ƒ₃(X, MA1; Khomani), with modern populations and selected ancient samples (DevilsGate1, DevilsGate2, Ust’-Ishim, Kotias, Loschbour, and Stuttgart), using all SNPs, with ƒ₃ > 0.15 displayed.
  
• fig. S12. D scores of the form D(X, Khomani; MA1, DevilsGate1), with all modern populations in our panel and selected ancient samples, using all SNPs.
  
• fig. S13. D scores of the form D(X, Khomani; MA1, DevilsGate1), with all modern populations in our panel and selected ancient samples, using all SNPs.
  
• fig. S14. Outgroup ƒ₃ scores of the form ƒ₃(X, Ust’-Ishim; Khomani), with modern populations and selected ancient samples (MA1, Kotias, Loschbour, and Stuttgart), using all SNPs, with ƒ₃ > 0.15 displayed.
  
• fig. S15. D scores of the form D(X, Khomani; Ust’-Ishim, DevilsGate1), with all modern populations in our panel and selected ancient samples, using all SNPs.
  
• fig. S16. D scores of the form D(X, Khomani; Ust’-Ishim, DevilsGate2), with all modern populations in our panel and selected ancient samples, using all SNPs.
  
• fig. S17. Comparison of Devil’s Gate–related ancestry in the Ulchi and European hunter-gatherer–related ancestry in European populations.
  
• fig. S18. Comparison of Devil’s Gate–related ancestry in the Ulchi and Early European farmer–related ancestry in European populations.
  
• fig. S19. Comparison of Devil’s Gate–related ancestry in the Ulchi and Bronze Age Steppe–related ancestry in European populations.
  
• table S1. Details of sample preparation and sequencing.
  
• table S2. mtDNA contamination estimates.
  
• table S3. Admixture ƒ₃(Source1, Source2; Target) for the Ulchi with z < −1 using all SNPs.
  
• table S4. Admixture ƒ₃(Source1, Source2; Target) for the Ulchi with z < −1 using only transversion SNPs.
  
• table S5. Admixture ƒ₃(Source1, Source2; Target) for the Sardinians using all SNPs and showing the 10 most significantly negative pairs.
  
• table S6. Admixture ƒ₃(Source1, Source2; Target) for the Lithuanians using all SNPs and showing the 10 most significantly negative pairs.
  
• Legends for extended data figs. S1 to S8
  
• Legends for extended data tables S1 to S9
  
• References (55–84)

Download PDF

Other Supplementary Material for this manuscript includes the following:

• extended data fig. S1 (.pdf format). Results from ADMIXTURE analysis using the regional panel, all SNPs, and MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 10.
  
• extended data fig. S2 (.pdf format). Results from ADMIXTURE analysis using the regional panel, transversion SNPs, and MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 10.
  
• extended data fig. S3 (.pdf format). Results from ADMIXTURE analysis using the regional panel, all SNPs, and no MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 10.
  
• extended data fig. S4 (.pdf format). Results from ADMIXTURE analysis using the regional panel, transversion SNPs, and no MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 10.
  
• extended data fig. S5 (.pdf format). Results from ADMIXTURE analysis using the total panel, all SNPs, and MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 20.
  
• extended data fig. S6 (.pdf format). Results from ADMIXTURE analysis using the total panel, transversion SNPs, and MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 20.
  
• extended data fig. S7 (.pdf format). Results from ADMIXTURE analysis using the total panel, all SNPs, and no MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 20.
  
• extended data fig. S8 (.pdf format). Results from ADMIXTURE analysis using the total panel, transversion SNPs, and no MapDamage treatment on samples from Devil’s Gate and setting the number of clusters to K = 2 to 20.
  
• extended data table S1 (Microsoft Excel format). Sample information.
  
• extended data table S2 (Microsoft Excel format). ADMIXTURE proportions.
  
• extended data table S3 (Microsoft Excel format). Outgroup ƒ₃ statistics for Devil’s Gate.
  
• extended data table S4 (Microsoft Excel format). Outgroup ƒ₃ and space.
  
• extended data table S5 (Microsoft Excel format). Outgroup ƒ₃ for MA1 and Ust’-Ishim.
  
• extended data table S6 (Microsoft Excel format). D scores for MA1 and Ust’-Ishim.
  
• extended data table S7 (Microsoft Excel format). D scores for the Ulchi.
  
• extended data table S8. Admixture ƒ₃ for the Koreans and the Japanese.
  
• extended data table S9 (Microsoft Excel format). Phenotypes of interest.

Download Extended Figures

Download Extended Data Tables