Biplot with main effect variables. Poor case/control distinction

Legend: Figure (biplot) shows at the same time the gene loadings (in green) and subjects scores (blue controls and red cases). Genes having high loading 1 are noted, and a line is drawn from the origin. The case/control line shows the direction of the case/control distinction.

Statistical insight: A case/control dummy variable was added into the Eigen analysis. The case/control direction is not parallel with the first component and has a relatively small loading1, indicating poor overall separation of the cases and controls, when only the main effects of genes are analyzed.

Clinical insight: Investigating only genes main effect leads to poor overall separation of the cases and controls.

Biplot with main effect variables and gene interactions.

Legend: Gene by gene interaction is investigated. Figure (biplot) shows at the same time the gene and gene.x.gene loadings (in green) and subjects scores (blue controls and red cases). The case/control line shows the direction of the case/control distinction. Two other lines SJS and TEN indicate the direction of each phenotype.

Statistical insight: There is almost complete separation of cases from controls when two marker epistasis is allowed. Adding dummy markers for each syndrome (SJS, TEN) into the analysis suggests that two different genetic determinants broadly correlate with the two syndromes.

Clinical insight: Within the general genetic predisposition for this Type 2 idiosyncratic immune mediated epidermal adverse event there are two genetic epistatic subtypes.