The ROC curves for SVMs trained with different feature classes. The true positive rate (i.e., sensitivity) was plotted against the false positive rate (i.e., 1-specificity). (1). All three classes of features combined. (2). ORF only; (3). Conservation only; (4). Nucleotide sequence based features only.
Sun et al. BMC Genomics 2013 14(Suppl 2):S7 doi:10.1186/1471-2164-14-S2-S7