Briefly, GARFIELD tests for enrichment of GWAS associated variants in annotation categories. The user provides the GWAS P values for all variants and an annotation file indicating whether variants are located within functional categories to be tested. The software then selects an independent set of variants by sequentially removing variants with r2 > 0.1 within a 1-Mb window from the most significantly associated variant and then classifies each variant as overlapping a functional category if either the variant or a correlated variant (defined as r2 > 0.8) is part of the annotation category. Statistical significance is calculated using a generalized linear model, while variants are matched by MAF, distance to the nearest transcription start site, number of LD proxies and CpG and GC content. CpG and GC content were calculated for a region 500 bp up and downstream of the variant using the BSgenome.Hsapiens.UCSC.hg19 R package. All variants from the UK10K were used as the background set of genetic variants. Depending on the particular enrichment analysis, our eQTL data were either considered as the GWAS trait (using the minimum P
