Fine-mapping accuracy can be improved by integrating functional annotation data such as predicted regulatory elements from the ENCODE and ROADMAP Epigenomics projects55,56. This approach is motivated by early studies showing that disease-associated variants are systematically enriched in chromatin marks that delineate active regulatory regions in disease-relevant cell types57,58. Under this paradigm, a statistical model is developed to jointly estimate functional enrichment and update posterior probabilities of causality using functional annotations47,52,59,60. Some integrative methods assume that SNPs are unlinked60 or assume a single causal variant per locus52,59, but a recent study built upon the multiple causal variant model of ref.46 to incorporate functional annotation data47. In an analysis of rheumatoid arthritis summary association data, integrative fine-mapping using this approach reduced the average size of 90% credible sets by 10%61. In addition to increasing fine-mapping accuracy, these studies have also provided insights into polygenic architectures (see below) by identifying tissue-specific functional annotations that are enriched for causal disease signals. This can also be achieved by conducting fine-mapping without integrating functional annotation data (typically under a single causal variant assumption) and then overlapping
