Skip to main content
Figure 4 | BMC Bioinformatics

Figure 4

From: How to find simple and accurate rules for viral protease cleavage specificities

Figure 4

ROC for different rule based systems on the out-of-sample HIV-1 PR 3261 – detail. Detail of the receiver operator characteristic (ROC) for the OSRE rules' accuracy on the out-of-sample HIV-1 PR 3261 data set. The squares and the crosses are the results for Table 4 and Fig. 1a in [17]. The squares show the results when using 1, 2, 3, ..., and 10 rules from Fig. 1a in [17] and the crosses for Table 4 in [17]. The dots show the prediction accuracy for the HIVcleave web-server [37]. The solid curve shows the ROC curve for the hitherto best predictor, a linear support vector machine trained on the HIV-1 PR 1625 data set [38], which does not provide any rules. The dashed diagonal line marks the expected results for random prediction.

Back to article page