Conducting genome-wide association studies (GWAS) Epistasis scenarios (chapter 6)
In general, genome-wide association studies (GWAS) apply univariate statistical tests to each gene marker or single nucleotide polymorphism (SNP) as an initial step. This SNP-based test is statistically straightforward, and the core tests for assessing the associations are standard methods (e.g., χ2 tests, regression) that have been studied outside of the GWAS context. Kuo and Feingold describe the most commonly used statistical methods that are applied to GWAS.2 All tests cited in the chapter are single-locus tests. If the genetic inheritance properties are not known, we recommend combining two or more statistical tests.3 In many cases, the SNPs associated with a disease are not located in a region of DNA that codes for a protein. Instead, they are located in the large noncoding regions between genes or in intron sequences, which are edited out of mRNAs prior to translation to proteins. These regions are presumably sequences of DNA that modify gene expression, but usually their functions are unknown.
Cooley, P., Gaddis, N., Folsom Jr., R. E., & Wagener, D. (2016). Conducting genome-wide association studies (GWAS): Epistasis scenarios (chapter 6). In P. Cooley (Ed.), Methods in statistical genomics: In the context of genome-wide association studies (pp. 65-84). (RTI Press Publication No. BK-0016-1608). Research Triangle Park, NC: RTI Press. https://doi.org/10.3768/rtipress.2016.bk.0016.1608