Data from Rossen et al. 2026 medRxiv, "Per-allele disease and complex trait effect sizes are predominantly African MAF-dependent in European populations"

Data from Rossen et al. 2026 medRxiv, "Per-allele disease and complex trait effect sizes are predominantly African MAF-dependent in European populations". https://www.medrxiv.org/content/10.64898/2025.12.31.25343290v2 maf_censored.tsv.gz contains 3 allele frequency measurements. MAF_afr70346 contains allele frequencies for 70,346 African-ancestry individuals from All of Us v8 (assigned using continental ancestry groups defined by All of Us). MAF_eur70346 contains allele frequencies for 70,346 European-ancestry individuals from All of Us v8 (assigned using continental ancestry groups defined by All of Us). MAF_afr_unadmixed contains unadmixed African allele frequencies, estimated by restricting to African individuals with two African haplotypes locally (this computation was restricted to 46,672 African ancestry individuals who were present in All of Us v7; see methods section of manuscript for further details). Note the following: We restricted to variants with either African MAF > 0.001 or European MAF > 0.001, and missingness < 0.05 in both European and African ancestries, according to the All of Us Variant Annotation Table. Allele frequency calls corresponding to MAC <= 40 have been censored as required by the All of Us Data and Statistics Dissemination Policy. All NaN values correspond to SNPs with MAC <= 40. The MAF column is computed by plinkv1.9. This means that the “MAF” can be greater than 0.5. Afr70,346 is not actually used in the manuscript. This data is not restricted to single nucleotide polymorphisms, but the analyses in the manuscript do restrict to single nucleotide polymorphisms. Variants are called against hg38 reference.