When the appropriate datasets are available, computational approaches can offer a relatively fast and cost-effective way of identifying putative enhancer-gene interactions in a given cell type. However, they are generally limited to detecting a subset of enhancer-promoter interactions within a given cell type, and none are capable of identifying trans interactions. Methods that rely on cell-type specificity or concordant changes in enhancers and genes across cell types may lack the sensitivity to predict interactions for ubiquitously expressed genes or to delineate interactions in domains with a high density of cell-type-specific genes. There is no standard or ‘reference’ dataset to validate the accuracy of gene-enhancer predictions. Thus, each study utilizes a different approach to evaluate accuracy, which makes it difficult to determine which method is most accurate. This necessitates experimental validation of enhancer-gene interactions determined using prediction-based methods. Despite these limitations, computational approaches can help to identify the targets of enhancer-risk variants. The method developed by Thurman and colleagues was applied to all GWAS loci and predicted gene targets of 419 disease-associated risk variants [20], most of which were located more
