Global skin colour prediction from DNA

• DOI: 10.1007/s00439-017-1808-5
• Publication Date: 2017-05-12T14:52:41Z
• Authors: Susan Walsh; Lakshmi Chaitanya; Krystal Breslin; Charanya Muralidh...; Agnieszka Broniko...; Ewelina Pospiech; Julia Koller; Leda Kovatsi; Andreas Wollstein; Wojciech Branicki; Fan Liu; Manfred Kayser
• Keywords: SNP

Human skin colour is highly heritable and externally visible with relevance in medical, forensic, anthropological genetics. Although eye hair can already be predicted high accuracies from small sets of carefully selected DNA markers, knowledge about the genetic predictability limited. Here, we investigate predictive value 77 single-nucleotide polymorphisms (SNPs) 37 loci previously associated human pigmentation using 2025 individuals 31 global populations. We identified a minimal set 36 informative SNPs developed statistical prediction model capable on scale. Average cross-validated expressed as area under receiver-operating characteristic curve (AUC) ± standard deviation were 0.97 0.02 for Light, 0.83 0.11 Dark, 0.96 0.03 Dark-Black. When 5-category, this resulted 0.74 0.05 Very Pale, 0.72 0.73 Intermediate, 0.87±0.1 A comparative analysis 194 independent samples 17 populations demonstrated that our outperformed proposed 10-SNP-classifier approach AUCs rising 0.79 to 0.82 White, comparable at intermediate level 0.63 0.62, respectively, large increase 0.64 0.92 Black. Overall, study demonstrates chosen markers model, particularly 5-category level; allow predictions within between continental regions first time, which will serve valuable resource future applications forensic anthropologic