Fig. 6

Distribution of AUC scores obtained from 1000 sets of randomly selected 9 SNPs. Distribution of AUC-ROC scores binned in intervals of 0.01 of 1000 sets of randomly selected 9 SNPs to verify that predictive performance observed from the selected 9 SNPs by the feature selection pipeline employed was a non-random occurrence. AUC-ROC scores were obtained by evaluation in each of the three unseen test sets using one of the five chosen ML classifiers, the Random Forest classifier