Table 1. Gold standard independent DNA-methylation prevalence estimation, and sensitivity and specificity estimation for HM450 and MethylCap-seq.
| Beta | Stat. | Method | Type 1 chemistry loci | Type 2 chemistry loci | ||||
|---|---|---|---|---|---|---|---|---|
| Mean (SD) | Min. | Max. | Mean (SD) | Min. | Max. | |||
| 0.2 | Prev. | – | 54.0 (3.3) | 48 | 60.9 | 79.5 (1.5) | 75.6 | 83.0 |
| Sens. | HM450 | 76.8 (2.1) | 71.6 | 80.9 | 91.3 (0.9) | 89.3 | 93.6 | |
| Spec. | HM450 | 85.7 (1.0) | 83 | 87.6 | 80.3 (0.8) | 78.3 | 82.1 | |
| Sens. | MCap | 50.1 (9.2) | 24.7 | 67.1 | 32.8 (8.7) | 14.1 | 52.0 | |
| Spec. | MCap | 94.1 (1.3) | 90.6 | 97.1 | 93.4 (1.1) | 90.9 | 95.3 | |
| 0.3 | Prev. | – | 51.5 (3.5) | 44.5 | 60.0 | 75.3 (2.1) | 69.6 | 78.5 |
| Sens. | HM450 | 74.3 (2.5) | 66.5 | 79.0 | 89.5 (1.2) | 87.2 | 92.1 | |
| Spec. | HM450 | 86.8 (1.1) | 84.5 | 89.5 | 81.0 (0.9) | 79.2 | 83.7 | |
| Sens. | MCap | 52.5 (9.0) | 26.0 | 68.1 | 34.5 (9.3) | 15.0 | 57.6 | |
| Spec. | MCap | 93.8 (1.4) | 90.5 | 96.9 | 93.1 (1.4) | 89.5 | 96.2 | |
| 0.4 | Prev. | – | 49.4 (3.9) | 41.1 | 59.1 | 71.2 (2.9) | 63.0 | 77.5 |
| Sens. | HM450 | 71.8 (2.6) | 65.2 | 77.5 | 87.8 (1.4) | 84.2 | 90.9 | |
| Spec. | HM450 | 87.7 (1.2) | 85.4 | 90.4 | 81.6 (0.9) | 79.7 | 83.8 | |
| Sens. | MCap | 53.9 (8.5) | 27.1 | 68.5 | 35.8 (9.4) | 15.6 | 59.8 | |
| Spec. | MCap | 93.1 (1.7) | 88.8 | 96.7 | 92.8 (1.5) | 89.3 | 95.7 | |
Data indicate mean (standard deviation, SD) and minimum (min.) and maximum (max.), over the different samples, for different statistics (stat.) for both methods (MCap indicates MethylCap-seq): methylation prevalence (prev.), sensitivity (sens.) and specificity (spec.) for both chemistries and different HM450 beta-value thresholds (Beta) for presence of methylation. Models presenting strong convergence problem and two samples with very low coverage (samples 38 and 43, Supplementary Table 1) were not considered in the estimation of the statistics. Nevertheless, for each statistic at least 90% of the samples were used.