Fig. 7: Inferring the relationship between archaic and modern human facial shape using face-associated SNPs.

a Regional-PVE effects of lead SNPs in Neanderthal introgressed segments on the face, divided into EUR-specific, overlapping, and EAS-specific groups. b Top gene regions contributing to total effects for the three SNP groups in a, showing cumulative contributions accounting for 50% of the total. c Violin plots of squared correlations between facial PRS profiles of archaic humans and four major populations in the 1000 Genomes Project (African, AFR, N = 504; European, EUR, N = 504; East Asian, EAS, N = 503; South Asian, SAS, N = 489), grouped by Neanderthals, Neanderthal-Denisovan admixed individual, and Denisovan. The band indicates the median, the box indicates the first and third quartiles, and the whiskers indicate 1.5 times of interquartile range from box. d The mean absolute frequency differences between 10 archaic human samples and 2000 samples from four major populations in the 1000 Genomes Project (same as c) were illustrated for different sets of SNPs, with higher values indicating greater dissimilarity (see the “Methods” section). The dashed line corresponds to the 357 SNPs used to construct the archaic human facial PRS profiles. The violin plots represent replicates of 10,000 random sets of SNPs. The band indicates the median, the box indicates the first and third quartiles, and the whiskers indicate 1.5 times of interquartile range from the box. Labels on the x-axis denote the different SNP sets used: “ALL” represents random 357 genome-wide SNPs, “EUR segments” and “EAS segments” include the SNPs among the 357 overlapping introgression segments specific to Europeans and East Asians, respectively. The 3D template facial image in this figure is adapted from White et al.⁸² published under an Open Access license (CC BY 4.0), see http://creativecommons.org/licenses/by/4.0/.