Skip to main content

Table 1 Logit model generation

From: Do little interactions get lost in dark random forests?

Interaction model

SNP1

SNP2

SNP1 x SNP2

Interaction only

0

0

β I

Modifier SNP

0

β I

β I

No interaction

β I

β I

0

Redundant

β I

β I

β I

Synergistic

β I

β I

β I

  1. In the simulation studies, 2 interacting SNPs and several SNPs having only marginal effects or no effects (noise SNPs) were generated. The phenotypes were simulated with additive effects and logit models. The interacting SNPs have marginal and/or interaction effects, depending on the genetic model. All effects of the interacting SNPs are generated from a single coefficient β I . The table shows marginal effects of SNP1 and SNP2 and the interaction effect. If variants at both SNPs are present, the resulting effect is the sum of the marginal effects and the interaction effect. The baseline β 0 was chosen to generate an approximate equal number of cases and controls for each scenario