Lack of epistatic interaction between the SLE-associated SNP in TYK2 and IRF5.
| SNP | No. of Subjects | Models with or without Interaction* | LRASSOC Models ** | |||||
|---|---|---|---|---|---|---|---|---|
| TYK2 | IRF5 | SLE | Controls | Best Model | p | Best Model | AIC to Interaction | |
| Hellquist2 | rs2304256 | rs10954213 | 277 | 356 | With interaction | 0.014 | – | – |
| Our data3,8 | rs2304256 | rs10954213 | 419 | 454 | No difference | 0.68 | Additive† | 1.83 |
| rs2304256 | rs10488631 | 1223 | 1300 | No difference | 0.3 | Additive | 1.98 | |
| rs2304256 | rs729302 | 1223 | 1300 | No difference | 0.88 | Additive | 0.94 | |
↵* Multivariate logistic regression models with and without an interaction term were compared as in Hellquist, et al2. The models did not specify any mode of inheritance. P values were obtained with the likelihood ratio test.
↵** Comparison of the specific inheritance models included in LRASSOC9. Akaike’s Information Criterion (AIC) was used to classify the models. A value of AIC > 2 indicates a meaningful difference.
↵† Additive: model with additive effects of each SNP, without dominance or interactive components. SNP: single-nucleotide polymorphisms.