Gene-based genome-wide association analyses (GWGAS) were conducted in MAGMA (http://ctg.cncr.nl/software/magma23), using the P values from the GWA analyses as input. This gene-based analysis tests the joint signal of all SNPs in a gene with the phenotype, while accounting for LD between those SNPs, thus uncovering gene-level signal that may go unnoticed in SNP-based analyses. In total, 18,183 genes were covered by at least one SNP, and used in the GWGAS. Within each gene-based analysis, we applied a stringent Bonferroni correction for the number of tested genes, resulting in a genome-wide threshold for significance of P < 2.75 × 10−6 (0.05/number of genes tested). In addition, we indicate whether the gene-signal survives correction for the 13 phenotypes tested (P < 2.12 × 10−7; Supplementary Data 33; Supplementary Fig. 22).
