Table
3: Tool
performance comparisons.
Category |
Tool\Metric |
TPR |
PPV |
F-measure |
MCC |
|
Single sequence methods |
Our algorithms |
SSD-liberal |
0.754 |
0.847 |
0.798 |
0.799 |
SSD-opt |
0.662 |
0.811 |
0.729 |
0.732 |
||
nbRSSP-extractor |
0.558 |
0.441 |
0.493 |
0.496 |
||
RNALfold-lnrz |
0.461 |
0.473 |
0.467 |
0.467 |
||
MFE-based |
M fold |
0.52 |
0.541 |
0.53 |
0.531 |
|
RNA fold |
0.48 |
0.468 |
0.474 |
0.474 |
||
Fold |
0.658 |
0.594 |
0.624 |
0.625 |
||
ML-based |
Context Fold |
0.780 |
0.787 |
0.784 |
0.784 |
|
MEA-based |
IPknot |
0.580 |
0.676 |
0.624 |
0.626 |
|
Centroid Fold |
0.550 |
0.717 |
0.622 |
0.628 |
||
S fold |
0.503 |
0.508 |
0.505 |
0.505 |
||
Fold then align |
MXScarna |
0.610 |
0.725 |
0.663 |
0.665 |
|
Comparative approaches |
MARNA |
0.506 |
0.729 |
0.597 |
0.608 |
|
Align then fold |
CentroidAlifold |
0.650 |
0.867 |
0.743 |
0.751 |
|
RNAalifold |
0.707 |
0.781 |
0.742 |
0.743 |
||
P fold |
0.400 |
0.810 |
0.536 |
0.569 |
||
Fold and
align |
Dyn align |
0.719 |
0.844 |
0.776 |
0.779 |
|
Fold align |
0.615 |
0.293 |
0.397 |
0.425 |
||
Carnac |
0.571 |
0.871 |
0.69 |
0.705 |
||
Base
pairing probability |
Turbo Fold |
0.790 |
0.747 |
0.768 |
0.768 |
|
RNA sampler |
0.692 |
0.904 |
0.784 |
0.791 |
Metrics
over different RNA datasets (e.g., RNA strand, LSU, RNAse
p and SM-A05), TPR = True Averaged rRNA, SSU, LGW17 positive rate; PPV = Positive Predictive Value; F-Measure, ;
MFE = Minimum Free Energy; MEA = Maximum Speed Accuracy; ML = Machine Learning.