Another limitation of our method is that its results may be difficult to validate. We undertook a type of validation using independent chromatin data, when there was comparable chromatin data available. However, this type of validation involves a number of challenges. First, we often do not have chromatin data in the same tissues and cell types as the gene expression data. Second, it is not clear that we should always expect results to replicate; for example, it is biologically plausible that SNPs near specifically expressed genes in the relevant tissue are enriched, while SNPs in H3K36me3 peaks called in the tissue are not. Third, our gene expression annotations represent relative activity—we select genes with higher expression in the focal tissue compared to other tissues—while the chromatin annotations that we use here represent absolute activity (although relative chromatin annotations are also possible6,66). Despite these limitations, replicating an enrichment for a particular system, tissue, or cell type using independent chromatin data can provide a strong validation for gene expression results.
