This vignette comprehensive instructions to obtaining de novo differentially-methylated regions (DMRs) between the user-defined sample groups using DiMmer(1), following a successful CpG-level differential methylation analysis with RnBeads for Infinium HumanMethylation450 BeadChip and Infinium MethylationEPIC v1.0 datasets.

Introduction

Differentially-methylated regions (DMRs) are contiguous regions of the genome exhibiting context-specific differential methylation. DiMmer facilitates de novo identification of such regions. Prior to employing DiMmer, it is essential to conduct a CpG-level differential methylation analysis using RnBeads.

Prerequistes

1. Java installation: Ensure that a suitable version of Java and Java Development Kit (JDK) 11 or above is installed on your system.

2. Running RnBeads: The de novo differentially-methylated region search requires CpG site-level differential methylation results from RnBeads. Detailed instructions on executing RnBeads for your methylome data can be accessed here.

Download and run DiMmer

After successfully running the differential methylation analyses for all desired sample group comparisons using RnBeads,

  1. Create a sub-directory within the RnBeads results directory titled ‘dimmer’.
  2. Subsequently, download the dimmer.jar file, dimmer_console.config file for DiMmer v2.3 from here and this R script, and move them into the ‘dimmer’ directory.
  3. Execute the script in R/Rstudio with the following command:

source('./dmrsearch.R')

NOTE: DMR search parameters used by DiMmer are defined in dmrsearch.R. If required, these parameters can be altered within the ‘config’ variable in this script.

Example outcome

This workflow was utilized to identify de novo DMRs between induced pluripotent stem cell (iPSC) and non-iPSC samples from a Infinium HumanMethylation450 BeadChip dataset with Gene expression omnibus (GEO) accession number—GSE76372(2).

Based on differential methylation of individual CpGs between sample groups, DiMmer identifies de novo DMRs and generates dimmer.csv containing scores and p-values for each DMR (Figure 1)1.

Figure 1: Scatterplot depicting relationship between the output scores (x-axis) and -log10(p-values) (y-axis) for the 5811 de novo DMRs identified by DiMmer between iPSC vs. non-iPSC samples from a 450K methylation array dataset(2). p-value of 0.05 is indicated by the horizontal red line.

References

  1. Frisch T, Gøttcke J, Röttger R, Tan Q, Baumbach J. DiMmer: Discovery of Differentially Methylated Regions in Epigenome-Wide Association Study (EWAS) Data. Methods Mol Biol. 2018;1807:51-62. doi: 10.1007/978-1-4939-8561-6_5. PMID: 30030803.

  2. Kitayama S, Zhang R, Liu TY, Ueda N, Iriguchi S, Yasui Y, Kawai Y, Tatsumi M, Hirai N, Mizoro Y, Iwama T, Watanabe A, Nakanishi M, Kuzushima K, Uemura Y, Kaneko S. Cellular Adjuvant Properties, Direct Cytotoxicity of Re-differentiated Vα24 Invariant NKT-like Cells from Human Induced Pluripotent Stem Cells. Stem Cell Reports. 2016 Feb 9;6(2):213-27. doi: 10.1016/j.stemcr.2016.01.005. PMID: 26862702; PMCID: PMC4750166.