Cindy Boer

262 | Chapter 6 gene. If the causal SNV is located in a coding region of a gene and affects protein transla- tion, folding or function it is likely that that gene would also be the causal gene. However, many of these variants are located in the non-coding regions of the genome. It is vastly more difficult to predict the possible biological consequences of these non-coding vari- ants than it is for coding variants, where much more is known about the consequences of sequence variation[25]. Yet, even non-coding causal SNVs can help identification of the causal gene. Elucidating GWAS: Causal Gene(s) GWAS associated SNVs are more likely to reside in gene regulatory regions, and thus are thought to exert their disease risk via affecting gene expression[26]. However, identify- ing which gene or genes might be targeted is not straightforward. It is a critical mistake to assume that noncoding variations, located intronic in a gene sequence will exert their function on that same gene, since only 14%of SNVs in non-coding regions target nearest genes[27]. Causal SNVs can be located as far as 2MB away from their target gene[28], due to the 3D structure of DNA, which connects DNA elements that are apparently far away from each other. Even larger distances between regulatory region and gene are possible as even larger topologically associated domains (TADs), have been observed [29]. These are large (megabase scale) genomic regions of interconnectivity, in which enhancers usually contact to genes also located within the TAD, but not to genes outside of the TAD. Fortunately, public databases exist of the 3D chromatin structure of multiple cell types and tissues to help identify such gene regulatory region - gene DNA loops. In chapter 3.1 and 3.2 evidence for such long-range interactions of ~200-700 kb between regulatory regions and target gene has been presented: rs10948155 with RUNX2 (~700 Kb distance) and rs10916199 with WNT9A (~200 Kb distance) . Elucidating GWAS: Causal Cell Type, Fate or State Finally, all annotation and functional genomic data should be examined in the context of cell types or cell states associated with the GWAS phenotype. Here, endophenotypes or stratified phenotypes can directly indicate the target cell type. For example, for mJSW, cartilage cells (chondrocytes) are the target cell type in which to examine the GWAS findings ( Chapter 2.1 ). In addition,, the GWAS associated SNVs, can indicate which cell types to examine. GWAS SNVs tend to be particularly enriched in enhancers that are ac- tive in the target cell types [26, 30]. By examining this enrichment for the osteoarthritis associated SNVs in the GO consortium, we identified osteoarthritis target tissues ( Chap- ter 4 ). These tissues included the known target tissues (cartilage and bone), but also in- dicated possible roles for muscle, endocrine and neurological tissues. Which in turn can be used to define novel osteoarthritis endophenotypes or stratified phenotypes, based