Table 3 Exhaustive search average variance explained by main effects and interaction effects.

From: Exhaustive search for epistatic effects on the human methylome

 

Discovery sample average variance explained

Replication sample average variance explained

All significant main effects

57.1%

57.8%

- Most-significant main effect

44.9%

45.4%

All significant interaction effects

8.2%

8.3%

- Most-significant LD-block associated effect

4.4%

4.5%

- Most-significant epistatic effects

7.8%

7.4%

All significant main effects and interaction effects

65.2%

66%

  1. The results are based on the N = 174 CpGs that showed at least one significant interaction effect when taking into account also main effects. For only 7 out of 174 CpGs (4%) no significant main effect of a SNP was detectable. 12 out of 174 CpGs (6.9%) showed both, a LD-block based effect as well as an epistatic effect. All significant main effects: average variance explained by all main effects that were kept in the final model. Most-significant main effect: average variance explained by the main effect that exhibited the smallest p-value. All significant interaction effects: average variance explained by all interaction effects that were kept in the final model; these were further separated in LD-block associated effects with SNP-pairs showing an r 2 > 0.021, or epistatic effects (r 2 ≤ 0.021), most-significant corresponds to the effect with the smallest p-value, if more than one of these effects were kept in the final model. All significant main effects and interaction effects: average variance explained by all main effects and interaction effects that were kept in the final model.
Back to article page