The ancestry deconvoluted UKBB admixed samples were then grouped according to their inferred ancestry fractions (Supplementary Table 2), and pPS calculated for height and BMI traits. We chose to explore only these two phenotypes due to the shortage of clearly labeled T2D or Breast Cancer cases in our admixed UKBB cohort, and because of the higher precision achievable in discerning genetic contribution to continuous traits. Similarly to the Egyptian, Ethiopian and ASW case, aspPS on the European component of the UKBB admixed samples recovered a distribution of scores within the European ranges. Conversely, PS on the non-European component of the UKBB samples shows non-zero median, as shown in Fig. 4a, b, (see also Supplementary Data 1). By correcting the population-level PS with the real phenotypic deviation, we were able to quantitatively describe this directional bias, which showed a suggestive correspondence with the FST distance between each Non-European group and our reference set of European descent, see Fig. 4c.
