There are several possible arbitrary ways to do so, including choosing the SNP closest to the transcription start site for each gene, choosing the most highly ranked SNP (based on the GWAS results) within a reasonable interval around each gene, or calculating a summary statistic of all the SNPs in an interval around each gene. These approaches, however, are likely to result in a bias towards enrichment of top-ranked pairs of SNPs and associated genes in genomic regions of high gene or SNP density or in regions characterized by high linkage disequilibrium (LD). As a result, the null expectation for the overlap of top-ranked pairs of SNPs and associated genes would be unclear.
