Figure 2
From: Assessing the druggability of protein-protein interactions by a supervised machine-learning method
ROC curves of the training data with the SVM model using all 69 PPI attributes (1:1 positives:negatives ratio). ROC curves with the linear (orange), polynomial (magenta), RBF (green), and sigmoid (blue) kernels were calculated for the 10,000 random training data sets, and average values of true positive rate at each false positive rate are plotted. AUCs ± standard deviations of the ROC curves with the linear, polynomial, RBF, and sigmoid kernels are 0.76 ± 0.09, 0.67 ± 0.20, 0.78 ± 0.13, and 0.64 ± 0.17, respectively.