Relative to recent association studies, the WTCCC has a small sample size (~2,000 cases and ~3,000 controls per disease). Thus, even with our novel method and a reduced multiple testing burden, our ability to detect numerous novel gene associations may be limited. Alternatively, we tested each disease for an enrichment of known disease genes identified from the NHGRI GWAS Catalog25. For each disease, we used the reported genes from the GWAS catalog as the set of known disease genes. We excluded studies listed in the NHGRI GWAS catalog that included the WTCCC samples in order to make sure our known gene lists were independent from the current analysis. We then counted the number of known disease genes that had a PrediXcan p-value below a given threshold. We compared this to the null expectation based on 10,000 randomly drawn gene sets of similar size to the known disease gene set to derive an enrichment p-value. We tested enrichment using PrediXcan p-value thresholds of 0.05 and 0.01.
