ROC curve of the combined score in the A. darlingi dataset. ROC curves for the cscore and HHMMiR in the A. darlingi dataset. The positive set consists of 44 clear homologs to A. gambiae pre-miRNAs. The negative sets consist of 1000 samples, each of the size of the positive set, drawn from the non-overlapping candidates. The dashed lines are the individual ROC curves for each sample. The solid lines are the average ROC curves. The average AUC (area under the curve) values are also shown. The green diamond represents the average performance of the triplet-SVM pre-miRNA classifier and the smaller green circles represent its performance on each sample.
Mendes et al. BMC Genomics 2010 11:529 doi:10.1186/1471-2164-11-529