Table 2.
Performance of methylation prediction using whole-genome bisulfite sequencing data
| Training set | Test set | Accuracy (%) | Precision (%) | Recall (%) | TN | FN | TP | FP | R | RMSE |
|---|---|---|---|---|---|---|---|---|---|---|
| WGBS 450K sites | WGBS non 450K sites | 92.4 | 86.5 | 44.5 | 78264 | 6095 | 4890 | 764 | 0.64 | 0.24 |
| WGBS 450K sites | WGBS 450K sites | 90 | 91.8 | 82.3 | 51260 | 6374 | 29726 | 2653 | 0.86 | 0.23 |
| 450K data | WGBS non 450K sites | 92.2 | 88.5 | 41.4 | 78437 | 6442 | 4543 | 591 | 0.62 | 0.23 |
| 450K data | WGBS 450K sites | 89.3 | 93.0 | 79.3 | 51762 | 7465 | 28635 | 2151 | 0.84 | 0.24 |
FN, false negatives; FP, false positives; RMSE, root-mean-square error; TN, true negatives; TP, true positives; WGBS, whole-genome bisulfite sequencing.