We showed that in admixed individuals aspPS can recover an unbiased distribution of PS by masking out the spurious shifting effect introduced by genomic segments derived from deeply divergent populations. This enables the assessment of a sample’s PS within a European reference set when dealing with PS calculated using European genetic effect sizes (Figs. 2, 4). The aspPS approach can be transferred to samples where a fraction of the genome is simply missing (pPS). On the other hand our approach allows the evaluation of the pPS predictivity (Fig. 3) on an individual-basis, hence enabling a per-sample evaluation of the reliability of the phenotype prediction. Our simulations and analyses on real admixed individuals from the UKBB show that, depending on the cross-population transferability of the adopted trait-SNP associations, the aspPS we introduced here provides an alternative mean of assessing an admixed individual’s PS and, when multiple aspPS are combined in a casPS, the casPS predictivity outperforms both aspPS and at least one of the total PS scores (European or East Asian total PS, in our examples). Intuitively, and according to our
