Fig. 1

Boxplots for predictive AUC of genotype models. Boxplots for the predictive AUC obtained for each fold in the 10-fold cross-validation (cv) approach used to learn genotype models from data under different algorithms: a Results for bipolar disorder (BD), b coronary artery disease (CAD), c hypertension (HT), d irritable bowel disease (IBD), e rheumatoid arthritis (RA), f Type 1 diabetes (T1D) and g Type 2 diabetes (T2D). For each plot (with the exception of hypertension where data was unknown), three horizontal lines are also plotted corresponding to AUC _quar (red line), AUC _half (green line) and AUC _max (blue line) [13]