Table 5.
Somatic SV detection performance on paired normal-tumor HCC1395 breast cancer data using PacBio and ONT sequencing
| Library | SV type | Metric | FocalSV (target) | FocalSV (auto) | SVIM-asm | cuteSV | SVIM | pbsv | Sniffles2 |
|---|---|---|---|---|---|---|---|---|---|
| HCC1395 PacBio | TRA | Benchmark | 137 | 137 | 137 | 137 | 137 | 137 | 137 |
| Comp | 119 | 74 | 25 | 30 | 58 | 452 | 173 | ||
| TP | 91 | 58 | 11 | 14 | 49 | 73 | 41 | ||
| FP | 28 | 16 | 14 | 16 | 9 | 379 | 132 | ||
| FN | 46 | 79 | 126 | 123 | 88 | 64 | 96 | ||
| Recall | 66.42% | 42.34% | 8.03% | 10.22% | 35.77% | 53.28% | 29.93% | ||
| Precision | 76.47% | 78.38% | 44.00% | 46.67% | 84.48% | 16.15% | 23.70% | ||
| F1 | 71.09% | 54.98% | 13.58% | 16.77% | 50.26% | 24.79% | 26.45% | ||
| INV | Benchmark | 133 | 133 | 133 | 133 | 133 | 133 | 133 | |
| Comp | 104 | 49 | 19 | 101 | 27 | 62 | 88 | ||
| TP | 55 | 30 | 8 | 9 | 23 | 31 | 29 | ||
| FP | 49 | 19 | 11 | 92 | 4 | 31 | 59 | ||
| FN | 78 | 103 | 125 | 124 | 110 | 102 | 104 | ||
| Recall | 41.35% | 22.56% | 6.02% | 6.77% | 17.29% | 23.31% | 21.80% | ||
| Precision | 52.88% | 61.22% | 42.11% | 8.91% | 85.19% | 50.00% | 32.95% | ||
| F1 | 46.41% | 32.97% | 10.53% | 7.69% | 28.75% | 31.79% | 26.24% | ||
| DUP | Benchmark | 230 | 230 | 230 | 230 | 230 | 230 | 230 | |
| Comp | 311 | 130 | 12 | 83 | 111 | 897 | 100 | ||
| TP | 190 | 114 | 3 | 40 | 77 | 99 | 92 | ||
| FP | 121 | 16 | 9 | 43 | 34 | 798 | 8 | ||
| FN | 40 | 116 | 227 | 190 | 153 | 131 | 138 | ||
| Recall | 82.61% | 49.57% | 1.30% | 17.39% | 33.48% | 43.04% | 40.00% | ||
| Precision | 61.09% | 87.69% | 25.00% | 48.19% | 69.37% | 11.04% | 92.00% | ||
| F1 | 70.24% | 63.33% | 2.48% | 25.56% | 45.16% | 17.57% | 55.76% | ||
| HCC1395 ONT | TRA | Benchmark | 137 | 137 | 137 | 137 | 137 | – | 137 |
| Comp | 99 | 146 | 346 | 13 | 4 | – | 1862 | ||
| TP | 82 | 57 | 10 | 8 | 3 | – | 43 | ||
| FP | 17 | 89 | 336 | 5 | 1 | – | 1819 | ||
| FN | 55 | 80 | 127 | 129 | 134 | – | 94 | ||
| Recall | 59.85% | 41.61% | 7.30% | 5.84% | 2.19% | – | 31.39% | ||
| Precision | 82.83% | 39.04% | 2.89% | 61.54% | 75.00% | – | 2.31% | ||
| F1 | 69.49% | 40.28% | 4.14% | 10.67% | 4.26% | – | 4.30% | ||
| INV | Benchmark | 133 | 133 | 133 | 133 | 133 | – | 133 | |
| Comp | 71 | 65 | 21 | 11 | 5 | – | 38 | ||
| TP | 49 | 20 | 3 | 2 | 2 | – | 26 | ||
| FP | 22 | 45 | 18 | 9 | 3 | – | 12 | ||
| FN | 84 | 113 | 130 | 131 | 131 | – | 107 | ||
| Recall | 36.84% | 15.04% | 2.26% | 1.50% | 1.50% | – | 19.55% | ||
| Precision | 69.01% | 30.77% | 14.29% | 18.18% | 40.00% | – | 68.42% | ||
| F1 | 48.04% | 20.20% | 3.90% | 2.78% | 2.90% | – | 30.41% | ||
| DUP | Benchmark | 230 | 230 | 230 | 230 | 230 | – | 230 | |
| Comp | 257 | 110 | 9 | 27 | 13 | – | 84 | ||
| TP | 168 | 93 | 2 | 11 | 9 | – | 76 | ||
| FP | 89 | 17 | 7 | 16 | 4 | – | 8 | ||
| FN | 62 | 137 | 228 | 219 | 221 | – | 154 | ||
| Recall | 73.04% | 40.43% | 0.87% | 4.78% | 3.91% | – | 33.04% | ||
| Precision | 65.37% | 84.55% | 22.22% | 40.74% | 69.23% | – | 90.48% | ||
| F1 | 68.99% | 54.71% | 1.67% | 8.56% | 7.41% | – | 48.41% |
-
The table summarizes the benchmark (ground truth set), comp (called set), TP (true positives), FP (false positives), FN (false negatives), recall, precision, and F1 scores for three SVs types—translocations (TRA), inversions (INV), and duplication (DUP)—evaluated across multiple benchmarked tools. The highest scores for recall, precision, and F1 are highlighted in bold.
