computed and the effective number of tests was 0.05/Pperm, reported in Supplementary Table 16. All phenotypes were shuffled together, so any phenotypic correlation was preserved. The effective number of tests corresponded to 88–95% of the total number of genes, indicating a small amount of statistical redundancy (Supplementary Note). To evaluate the TWAS approach, we computed phenotype association statistics for the 5,500 unrelated individuals and re-ran the analysis using only these summary statistics and the same expression reference panels. The resulting TWAS associations were nearly identical to the direct TWAS associations across the four traits (Pearson ρ=0.96). Reassuringly, the TWAS was generally more conservative than the direct estimates (Supplementary Figure 16).
