Strategies for Genetic Studies of Complex Diseases

In light of this, it is likely that GWAS hits found in intergenic regions far from known genes are true associations whose biology is not yet understood, rather than false positives. The human genome is incompletely annotated. Regions where GWAS associations have been found, but no known genes are located, could easily harbor unidentified new genes or regulatory elements. For instance, the authors point to the colon and prostate cancer risk SNP rs693267 located 335 kb upstream from the MYC gene on chromosome 8q24 (McClellan and King, 2010). This locus has been shown to physically interact with MYC and is associated with enhanced Wnt signaling. Therefore, although the biology of this locus is not fully understood, it suggests a paradigm where intergenic disease-associated SNPs alter enhancer elements, either directly or through linkage disequilibrium, and therefore cause differential regulation of disease-related genes. This observation leads to a broader point: A lack of biological understanding of how these disease-associated variants are pathogenic does not mean that there is no biology to discover. Although our understanding of the mechanisms by which disease risk loci contribute to pathogenesis currently lags behind the pace at which new loci are discovered, promising stories continue to emerge. To continue the previous example, although no definitive correlation between the rs6983267 genotype and MYC expression has been demonstrated, MYC is known to be tightly regulated and the right developmental time point may need to be examined to see such a correlation. Although further work is necessary to uncover the elusive mechanism by which the SNP confers risk, we propose that the existing evidence supports rather than refutes this SNP as a true cancer risk allele. Another example is a non-protein-coding region of chromosome 9q21 in which SNPs have been robustly associated with arterial disease. A recent paper reported that targeted deletion of an orthologous region in mouse interferes with cis-regulation of nearby genes (Cdkn2a/Cdkn2b) and may influence vascular cell proliferation (Visel et al., 2010). As a third example, an intronic type 2 diabetes risk SNP (rs7903146) was recently found to overlap with a region of islet cell-selective chromatin, and the two alleles of rs7903146 correlate with the open/closed chromatin state of the region (Gaulton et al., 2010). Thus, understanding the mechanisms by which GWAS loci contribute to disease will require considerable effort and time. We take this not as a sign that the common disease-common variant model has failed but rather that a challenge exists for the scientific community— a challenge that must be addressed with both traditional experimental genetics and innovative new approaches.