Abstract
Active regulatory elements within CD4+ T cells harbor disproportionate heritability (h2) for rheumatoid arthritis (RA). We hypothesized that regulatory elements specific to pathogenic CD4+ T cell-states better capture RA h2; however, defining these elements is challenging. To this end, we introduce IMPACT, a genome annotation strategy to identify cell-state-specific regulatory elements defined by key transcription factor binding profiles, learned from 398 chromatin and sequence annotations. Integrating IMPACT annotations of four CD4+ T cell-states with RA summary statistics of 38,242 Europeans and 22,515 East Asians, we observe that on average the top 5% of Treg predicted regulatory elements explain 85.7% (s.e. 19.4%, enrichment p<1.6e-05) of RA h2, and other cell-states explain a similar proportion. IMPACT captures RA h2 better than active CD4+ T cell regulatory elements, including super enhancers and specifically expressed genes (all p<0.05). IMPACT is generalizable to non-immune cell types and can identify other complex trait associated regulatory elements.