Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: TERIUS: accurate prediction of lncRNA via high-throughput sequencing data representing RNA-binding protein association

Fig. 2

Performance of the RPS classifier (a) Training and test ROC curves of the RPS. (b) Accuracy of the most-likely reading frame (MLRF). The difference between the CDS frame and MLRF for all human mRNAs (top) and those harboring a short CDS are shown separately (middle). The prediction using RNA-seq is shown at the bottom (random). (c) Benchmarking ORFscore. AUC values are 0.88 for both RPS and ORFscore. Dots and arrows indicate the cutoff used to assess performance throughout the paper. (d) Percent of mRNAs predicted as coding by the RPS and RiboTaper. The mRNAs were binned according to Ribo-seq RPM. (e) 2D plot of the RPS versus the ORFscore (left) and RiboTaper (right). Dotted grey lines are the RPS cutoff (0.5), RiboTaper cutoff (0.05), and 15th percentile of the ORFscore, calculated using all RefSeq protein-coding genes. 11 outliers (ORFscore > 1000) are excluded from the plot. NP written at the top-left corner indicates genes plotted in the quadrant are classified as ncRNA (N) by RPS and as protein-coding (P) by other methods, as plotted along the x-axis

Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us