Table 6 Result of different tools with different threshold values (used to determine when a called inversion matches benchmark)
From: InvBFM: finding genomic inversions from high-throughput sequence data based on feature mining
Thre-shold | Tool | No. Calls | TP | TP0 | FP | FN | Precision | Recall | F1-score |
|---|---|---|---|---|---|---|---|---|---|
ISPE | Delly | 1142 | 183 | 150 | 959 | 88 | 16.02% | 63.03% | 25.55% |
Lumpy | 66 | 51 | 51 | 15 | 187 | 77.27% | 21.42% | 33.55% | |
LumpyEP | 62 | 47 | 47 | 15 | 191 | 75.81% | 19.75% | 31.33% | |
Pindel | 649 | 84 | 79 | 565 | 159 | 12.94% | 33.19% | 18.62% | |
InvBFM | \( \frac{\mathbf{1379}}{\mathbf{1919}} \) | \( \frac{\mathbf{359}}{\mathbf{365}} \) | \( \frac{\mathbf{163}}{\mathbf{166}} \) | 1020 | 75 | 26.03% | 68.49% | 37.73% | |
ISPE *2 | Delly | 1142 | 244 | 164 | 898 | 74 | 21.37% | 68.91% | 32.62% |
Lumpy | 66 | 65 | 65 | 1 | 173 | 98.48% | 27.31% | 42.76% | |
LumpyEP | 62 | 60 | 60 | 2 | 178 | 96.77% | 25.21% | 40.00% | |
Pindel | 649 | 99 | 82 | 550 | 156 | 15.25% | 34.45% | 21.14% | |
InvBFM | \( \frac{\mathbf{1379}}{\mathbf{1919}} \) | \( \frac{\mathbf{462}}{\mathbf{468}} \) | \( \frac{\mathbf{167}}{\mathbf{172}} \) | 917 | 71 | 33.50% | 70.17% | 45.35% | |
ISPE *3 | Delly | 1142 | 258 | 165 | 884 | 73 | 22.59% | 69.33% | 34.08% |
Lumpy | 66 | 65 | 65 | 1 | 173 | 98.48% | 27.31% | 42.76% | |
LumpyEP | 62 | 61 | 61 | 1 | 177 | 98.39% | 25.63% | 40.67% | |
Pindel | 649 | 101 | 82 | 548 | 156 | 15.56% | 34.45% | 21.44% | |
InvBFM | \( \frac{\mathbf{1379}}{\mathbf{1919}} \) | \( \frac{\mathbf{479}}{\mathbf{485}} \) | \( \frac{\mathbf{170}}{\mathbf{173}} \) | 900 | 68 | 34.73% | 71.43% | 46.74% |