Supplementary materials.

Additional derivations for the Gibbs sampler. Table A. Linear relationship between the number of causal SNPs and heritability. We model the linear relationship between the number of causal SNPs for a trait and the heritability across all regions of the genome. We report the slope of the regression and the standard error. The slope can be interpreted as the expected per-SNP heritability contribution per causal SNP. The last column reports the number of ‘outlier’ regions for each trait, defined as a region with an absolute studentized residual greater than 3. Table B. Covariates that are associated with regional heritability . We perform a multivariate regression of heritability on the number of SNPs, number of causal SNPs, number of genes, median B-statistic, and non-cell-type-specific annotations [28]. Only the number of causal SNPs () remains significant for all traits after the multiple testing correction (average p-value = 6.37 × 10⁻¹¹), and the number of SNPs (M_r) remains significant for 3 our of 5 traits after the multiple testing correction. Table C. Likelihood ratio test assessing the role of gene density in regional polygenicity estimates. We perform a likelihood ratio test between the following two models to assess the effect of gene density on the number of causal SNPs after adjusting for both regional heritability and the number of SNPs .

(PDF)