Cindy Boer

74 | Chapter 2.1 Table 2: Association of mJSW loci with Hip osteoarthritis Hip osteoarthritis SNP Locus EA Beta SE p-value rs2862851 DOT1L A 0.06 0.02 6.9x10 -05 rs2236995 TGFA T -0.05 0.02 2.5x10 -03 SUPT3H-RUNX2 T 0.01 0.02 7.6x10 -01 rs10948155 SUPT3H-RUNX2 A -0.04 0.02 2.3x10 -02 PIK3R1 C 0.14 0.00 1.3x10 -04 rs496547 TREH-DOX6 A 0.01 0.02 5.8x10 -01 rs11880992 SLBP T -0.06 0.02 9.7x10 -05 SNP: single nucleotide polymorphism, EA: effect Allele, SE: standard Error. We observed the second variant in this genomic region, an intronic variant in RUNX2 , rs12206662, to have a larger effect size (β=0.14, p-value=1.1*10 −04 , r 2 =0.09 with rs10948172). We further examined whether the identified loci were found associated with other phenotypes in earlier reports ( Table 3 ). Five of the seven identified mJSW SNPs mapped to loci that have previously been associated with other bone-related pheno- types, primarily height. However, many of the identified height loci were not highly cor- related with the mJSW signal ( Table 3 ). Additional adjustment for height did not have an effect on the described association with mJSW; they showed an independent, possi- bly pleiotropic effect, on both traits. A particularly dense number of associations with different bone related phenotypes were present in the RUNX2 5’region, where variants have been associated to BMD[10], height[11], osteoarthritis[2] and ossification of the spine[12]. Given the low LD between the variants underlying the different GWAS sig- nals, it is likely that these represent independent associations. Prioritization of genes underlying the genetic loci We used multiple approaches that leverage different levels of information (e.g., gene expression, regulatory regions, published literature, mouse phenotypes) to prioritize candidate genes at each mJSW locus. Table 4 shows the summarized results from these analyses. In addition to the seven loci identified in the current study, we also analyzed five previously published loci for hip OA[2]. First, we focused on two gene prioritization methods: (i) DEPICT, a novel tool designed to identify the most likely causal gene in a given locus, and identify gene sets that are enriched in the genetic associations[21], and (ii) GRAIL which uses existing literature to identify connections between genes in the associated loci[22]. The DEPICT analysis yielded seventeen significantly prioritized