Distinct genetic architecture in the tails of complex traits

Key Points:

• The UK Biobank (UKB) genetic dataset includes 488,377 samples genotyped at 805,426 SNPs, with population ancestries defined via clustering and quality control steps resulting in 369,132 unrelated European ancestry individuals for primary analyses.
• An initial set of 408 continuous traits was filtered through rigorous phenotype and genotype quality control, resulting in 74 quantitative traits for primary analyses, residualized for environmental covariates and standardized.
• Polygenic risk scores (PRS) were calculated for these traits using GWAS in a base dataset and tested for deviations from common-variant architecture in trait tails using the POPout test, with replication performed in UKB repeated measures, multiancestry UKB samples, and the All of Us cohort.
• Sibling-based tests were applied to detect rare or de novo allele effects deviating from polygenic expectations, combining Mendelian and de novo effect tests into an omnibus statistic to assess departures from common-variant architecture.
• Analyses of rare variants from whole-genome and whole-exome sequencing data were integrated with common variant PRS to evaluate their contribution to trait variation and POPout effects, while forward simulations and lifetime reproductive success analyses were used to infer selection pressures on traits linked to genetic architecture.