Fig. 5: Using RNA expression to classify early-stage idiopathic PD status.
From: Early-stage idiopathic Parkinson’s disease is associated with reduced circular RNA expression
a Comparison of AUCs calculated in PPMI and ICICLE-PD cohorts using the expression of individual junctions (BSJ and FSJ) and BSJ:FSJ ratios to classify PD status. ROC curves were constructed using the expression of back-spliced junctions (BSJ, VST normalised), forward-spliced junctions (FSJs, VST normalised) and the circular to linear ratio (BSJ:FSJ ratio, calculated by CIRIquant) for each junction position. Spearman’s rho demonstrates the correlation between the AUC in each cohort. Grey dashed lines indicate an AUC of 0.5. b, c Ability of each predictor to classify PD status. ROC curves (b) showing the sensitivity and specificity of each predictor (see methods) at various thresholds in PPMI and ICICLE-PD cohorts. The area under the ROC curve for each predictor is shown in c. Performance in the PPMI was evaluated on held-out data in the outer nested folds. Performance in ICICLE-PD was based on models fit on the PPMI data. Error bars give the 95% confidence intervals of the AUC.
