Classifier Performance.

<p>The Receiver-operating characteristic (ROC) curve is shown for SVM-LIN, SVM-RBF, and SVM-seq (RBF classifier trained only on sequence data). ROC curves for each classifier showing false positive rate (fpr) and true positive rate (tpr), with the reference line for random classification is shown in gray. The difference between each classifier and the reference line shows the improvement over random of our method. The steep slope at the lower left of the classifier curves indicates that the highest-ranked predictions are most likely to be accurate for all three classifiers. Area under curve: SVM-LIN = 0.713, SVM-RBF = 0.734, SVM-seq = 0.563.