Interactions in Genetic Assocation Analysis

Statistical interaction occurs between two factors if their combined effect is different than what would be expected based on their individual separate effects. In genetics, differences in risk or mean phenotype between genotypes vary according to the exposure (interacting) variable.

Simple example: Phenylketonuria (PKU) is a genetic defect that causes severe intellectual disability only in the presence of dietary phenylalalnine. Every baby is tested for this trait to see if they need a special diet.

Modeling Statistical Interaction

We need a model to form an expectation for the joint effect, which we can use to define the interaction. The definition of interaction relies on some specification of non-interactive effects.

Let G_i be the genotype for person i, and E_i be an environmental factor for a person i

H₀: β_GE = 0; No interaction, the measure of association β_G between the distribution of the phenotype Y and the genotype G does not depend on E.