Figure 4From: Automatic annotation of protein motif function with Gene Ontology termsComparison of results with probability and M.I. A. ROC curves for classifying motif-term associations at different probability threshold. Filled circles are the results on the perfect match test set with an area under curve of 0.8715. Empty triangles are on the relaxed match test set with an area under curve of 0.871. Data points correspond to thresholds of p(Y = 1|X) from 0.9 to 0.1 (from left to right) with a step of 0.1. B. Precision (positive predictive value) at different probability cutoffs, where solid bars are the result on the perfect match test set and the open bars for the relaxed match test set. C. ROC curve for decision rules based on different M.I. cutoff thresholds with an area under curve of 0.816. D. Precision at different M.I. cutoffs.Back to article page