A regulatory DNA association study between autoimmune disease risk and variation in regulatory regions that are highly unique to adaptive immune cells. A. Madar1, D. Chang1, A. J. Sams1, F. Gao1, Y. Waldman1,2, C. Van Hout3, A. G. Clark1,3, A. Keinan1 1) Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, NY; 2) Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Tel Aviv, Israel; 3) Department of Molecular Biology and Genetics, Cornell University, Ithaca, NY.

   Autoimmune diseases arise from the abnormal response of adaptive immune cells (T- and B- lymphocytes) against body tissue. Variation in non-coding regulatory DNA is thought to be a major genetic contributor to multiple autoimmune diseases. Here, we integrate DNase-seq data (a high throughput technology to detect regulatory DNA marked by DNase1 hypersensitive sites, DHSs) from immune and other cell types, with genome-wide association studies of autoimmune diseases and, as controls, unrelated complex traits. We describe an approach that combines DNase-seq data from multiple cell types to quantify the level of specificity of DHSs to each cell type. Based on this new approach, we identified ~1.2 million nucleotides (0.04% of the genome) of regulatory DNA that is uniquely accessible in adaptive immune cell types. Our new quantitative approach greatly improves the detection of such regions compared to the more usual use of DHS data to infer a binary open/closed chromatin state. We show that these adaptive-immune-specific regulatory regions (but not regions specific to other cell types) selectively contribute to the risk of multiple autoimmune diseases, but not to other complex traits. For multiple sclerosis and rheumatoid arthritis, for instance, regulatory DNA that is most accessible in T regulatory cells is most associated with disease risk. Finally, we performed the first regulatory DNA association study (RDAS) of autoimmune diseases that considers only variants in or near adaptive-immune-specific regulatory regions, thereby reducing the multiple testing burden compared to GWAS. Associations that we discovered using a GWAS of a small number of individuals, and that were not discovered in the original GWAS, are highly replicable in more recent, larger studies, and lead to interpretable results for non-coding regulatory DNA. Our quantitative approach for detecting cell type-specific DHSs can generalize to many other applications. RDAS of trait-relevant cell types can facilitate new discoveries from GWAS, suggest new focus regions for trait specific array designs, and is particularly well tailored for the coming era of whole-genome sequencing based GWAS, as it can take advantage of the base-pair resolution offered by sequencing data, while avoiding the pitfalls of increasing the number of tests as a result of this resolution boost.

You may contact the first author (during and after the meeting) at