A novel statistical method — modeling genome-wide relatedness among 22,000 ancient and modern individuals and testing whether constant directional selection better predicts allele frequencies — combined with independent validation against UK Biobank trait-association data, shows that selection signals with high statistic values (above ~5) are overwhelmingly real rather than artifacts.

Reich explains how the team validated their selection statistic by showing it predicts enrichment for known trait-associated mutations from the UK Biobank — enrichment rises from 15% to roughly 60-70% as the statistic crosses a threshold of five, giving a calibrated way to know which signals are real. ✦ AI generated

David Reich · Dwarkesh Podcast · 2026-05-08 · original ↗

As we did that, the enrichment for genetic mutations that affect traits got higher and higher. Whereas it was only 15% when we didn’t use our selection statistic, when we required the selection statistic to be above about five, there was about a five-fold enrichment for mutations that cause traits.

verbatim transcript · starts at 114:10