Table 1.
Accuracy of various poly(A) prediction tools
| Tool | Results reported by authors | Results on our AATAAA dataset |
|---|---|---|
| Polyadq | MCC = 0.41–0.51 | Se = 28.23% |
| Sp = 83.88% | ||
| Acc = 56.05% | ||
| Polya_SVM (Cheng et al., 2006) | Se = 37.2–71.0% | Se = 58.30% |
| Sp = 74.6–96.7% | Sp = 64.42% | |
| Acc = 61.36% | ||
| Polyar | Se = 23.9–94.9% | Se = 57.28% |
| Sp = 14.7–66.4% | Sp = 49.69% | |
| Acc = 53.48% | ||
| Our Model (ANN) | Table 2 | Se = 80.55% |
| Sp = 83.57% | ||
| Acc = 82.06% | ||
| Our Model (RF) | Table 2 | Se = 86.10% |
| Sp = 91.60 | ||
| Acc = 88.90 | ||
| Polyah (Salamov, 1997) | MCC = 0.62 | |
| ERPIN | Se = 56% | |
| Sp = 69–85% | ||
| Polyapred | Se = 57.0% | |
| Sp = 75.8–95.7% | ||
| Poly(A) Signal Miner (Liu et al., 2003) | Se = 56.0–89.3% | |
| Sp = 67.5–93.3% | ||