paperKB
coga / coga-kb
Help
Sign in

Chunk #2 — Introduction

Source
Ancestry deconvolution and partial polygenic score can improve susceptibility predictions in recently admixed individuals.
Embedded
yes

Text

PS transferability has been proved exceptionally difficult across deeply divergent populations10–14,18,23, while ancestry deconvolution softwares display reliable performances in discerning genetic contributions from these groups of individuals22,24. Although human genetic diversity is distributed as a continuum across all continents, here we focus on the extremes of such a genetic gradient (i.e. genetic drift components that are modal in European, East Asian or Sub-Saharan African populations) to ease a preliminary exploration of the viability of our approach, and assume within population stratification as a lesser, yet important, source of bias. Hence, the drift components, or genetic ancestry labels, used here should just be seen as proxies for a broader gradient, not necessarily overlapping with cultural identity or self reported ethnicity of the analyzed individuals. Accordingly, we relied on a sample set including 120 Ethiopians, 100 Egyptians25 and 49 African-Americans26, all of which resulted from the admixture of African and West Eurasian populations approximately 100, 30, and 6 generations ago, respectively20,25. We also included 22054 samples from UK Biobank27, 5793 of which present an admixed genetic background(Supplementary Table 2). These samples display a polarized combination of highly studied and severely understudied genomic segments.