Table 2.
Comparison of numbers of regions identified by the two methods used for DMR and DMG detection
| Method used | # of CpG Regions | # of CpG regions with ≥ 60 CpGs | # of DMRs | # of DMRs with ≥ 60 CpGs | # of DMGs | Mean # of CpG sites per CpG Region | Mean # of CpG sites per DMR |
|---|---|---|---|---|---|---|---|
| SOMNiBUS (p < 6.05e−06; FWER < 0.05) | 8268 | 8268 | 131 | 131 | 125 | 108.3 | 124.5 |
| SOMNiBUS (p < 0.05) | 8268 | 8268 | 1228 | 1228 | 1183 | 108.3 | 113.2 |
| Bumphunter (q-value < 0.05) | 821,929 | 0 | 599 | 0 | 340 | 1.1 | 2.5 |
For SOMNiBUS, the FWER of 0.05 was estimated by a Bonferroni correction for the number of regions analyzed: 0.05/8268 = 6.05e−06. For bumphunter, the software uses permutations to estimate genome-wide false discovery rates, and [27] reported results for q-values < 0.05