the outcomes, with an equivalent phenotype across those with one or two copies of the risk allele) or recessive (two copies of the risk allele are needed before the phenotype is manifest, with individuals having 0 or 1 copy showing an equivalent phenotype) manner. These genetic models would necessitate a different coding scheme for the marker. Atheoretical approaches like GWAS traditionally model additive genetic influences, such that each copy of a specific variant is assumed to have the same magnitude of effect. In doing so, genotypes (e.g., AA/AG/GG) are coded 0, 1, or 2, with respect to a reference variant (in this case, G). In the absence of biologically plausible hypotheses and/or robust empirical evidence to the contrary, the analysis of candidate genes should generally start with an additive model. It would not be advisable, for instance, to arbitrarily collapse across genotypes solely for the purpose of increasing the statistical power of a model to detect significant associations. However, with the right justification, alternative coding schemes can be introduced. For example, if the literature on a specific SNP strongly suggests that the mere presence of a particular allele confers risk, the genotype may be coded to reflect the dominant influence
