Co-localization analysis was performed using the coloc package in R77. Cis-eQTL data for individuals of European ancestry from the GTEx v8 eQTL Tissue-Specific All SNP Gene Associations dataset (dbGaP Accession phs000424.v8.p2) were input as a quantitative trait into coloc (sample sizes for each tissue type indicated in Supplementary Table 15). Summary statistics from the gSEM analysis were input as a quantitative trait, with a sample size of 403,915. Summary statistics from the standard meta-analysis of OA were input as a case–control trait with 16,849 cases and 379,493 controls. All SNP positions were lifted over to build 38. The cis-eQTL data was partitioned into blocks based on the gene in the SNP-gene pair. For each gene block, only SNPs in the gSEM or meta-analysis summary statistics overlapping with the cis-eQTL data were input into the coloc function for (approximate) Bayes Factor colocalization analysis.
