Table 3.
Differentially methylated region detection method comparisons.
| Package | Apply to | Language | Algorithm | Usage | Note | Ref. |
|---|---|---|---|---|---|---|
| BSmooth | WGBS RRBS? |
R | Smooth/t test | DMR | Designed for WGBS, customization needed for RRBS; DMRs detected automatically; no covariates |
[66] |
| BiSeq | RRBS | R | Smooth/beta regression |
DMR | More specifically for targeted RRBS data; identify DMR (CpG cluster) automatically; allow covariates |
[67] |
| methylKit | WGBS RRBS |
R | Logistic regression |
DMR | Fisher’s test for a pair of samples and logistic regression for more samples with covariates; tiling window or predefined region for testing |
[61] |
| Fisher’s test | Annotation | |||||
| Bump hunting |
Array RRBS |
R | Linear regression |
DMR | Only for ratio data; allow covariates; auto DMR detection |
[68] |
| MOABS | BS data | C++ | Beta-binomial hierarchical model |
DMC/DMR | Group DMCs to DMR by a Hidden Markov Model |
[65] |
| Methylsig | RRBS WGBS |
R | Beta-binomial | DMC/DMR | Tiling window for DMRs (default 25 bps, likely too fragmented) |
[64] |
| Radmeth | WGBS RRBS |
C++ | Beta-binomial | DMC/DMR | Merge DMCs to DMR by weighted Z test for p-values |
[57] |
BS: Bisulfite sequencing; DMC: Differentially methylated CpG; DMR: Differentially methylated region; MOABS: Model-based analysis of bisulfite sequencing; RRBS: Reduced representation bisulfite sequencing; WGBS: Whole-genome bisulfite sequencing.