Plant DNA Methylation Data Analysis: CG/CHG/CHH Context, DMRs, and Stress-Response Validation
Plant DNA methylation analysis cannot be reduced to one methylation percentage or one list of differentially methylated genes. In plants, methylated cytosines occur in CG, CHG, and CHH sequence contexts, and the biological meaning, maintenance machinery, genomic distribution, and stress-response interpretation of each context can differ.
The practical challenge is to connect context-aware methylation calls to a defensible biological question. This guide outlines a research workflow from experimental design and WGBS or related sequencing through context-specific QC, DMR identification, gene and transposable-element annotation, RNA-seq integration, and targeted validation under stress conditions.
Figure 1: Plant methylation analysis should preserve sequence context and genomic annotation from the first QC table to the final candidate list.
Why CG, CHG, and CHH Must Be Analyzed Separately
In plant methylome studies, H denotes a base other than guanine. CG methylation is symmetric across the two DNA strands, whereas CHG and CHH methylation are non-CG contexts with different maintenance and targeting logic. The three contexts often occupy different genomic features and respond differently across tissues, developmental stages, genotypes, and stresses.
| Context | Common analytical interpretation | Main caution |
|---|---|---|
| CG | Stable methylation in gene bodies, promoters, and repetitive regions depending on species and tissue | A global CG average can hide locus-specific changes and cannot be compared across species without annotation context |
| CHG | Non-CG methylation associated with plant-specific maintenance and chromatin regulation | CHG calls are sensitive to sequence coverage, polymorphisms, and the definition of the local region |
| CHH | Often dynamic in regulatory or repetitive contexts and frequently relevant to RNA-directed DNA methylation | CHH is asymmetric and can be sparse; low coverage or poor conversion can create unstable estimates |
The presence of a DMR in one context does not automatically imply the same mechanism as a DMR in another. Context should therefore be retained in differential testing, annotation, visualization, and the final candidate table. A combined “methylated versus unmethylated” summary may be useful as a high-level overview, but it should not replace context-specific results.
For projects that need genome-wide coverage, the site’s Whole Genome Bisulfite Sequencing resource is a natural starting point. For lower-complexity or CpG-enriched designs, review Reduced Representation Bisulfite Sequencing with the plant genome and study question in mind.
Build the Experimental Design Before the Pipeline
Data analysis cannot rescue a design in which stress, genotype, tissue, and batch are inseparable. Before sequencing, define the biological unit, the treatment contrast, the time points, and the evidence required to call a stress-responsive methylation change.
At minimum, document:
- Biological factors: species or cultivar, genotype, tissue, developmental stage, sex where relevant, stress type, dose, duration, recovery period, and environmental conditions.
- Replicate structure: independent biological replicates per genotype and treatment, not only technical library repeats.
- Sampling logic: whether tissues are collected from matched positions, at the same circadian window, and before or after visible damage.
- Reference context: genome assembly, gene annotation, transposable-element annotation, SNP or haplotype resources, and the cytosine-calling convention used.
A robust design usually includes a matched untreated or well-watered control, the stress condition, and enough biological replicates to estimate within-group variation. If the objective is to compare tolerant and sensitive cultivars, genotype and treatment should be modeled as separate factors when the sample size supports it. If only a small pilot is possible, describe the result as candidate discovery rather than a general stress-response rule.
Choose the Methylation Assay for the Question
Whole-genome bisulfite sequencing provides broad single-base methylation coverage but can be demanding for repetitive plant genomes and low-input or degraded DNA. Enzymatic conversion approaches may reduce bisulfite-associated DNA damage in suitable workflows. Reduced-representation approaches can be cost-efficient but may not cover the regulatory or transposable-element regions that matter for a particular plant question.
| Research objective | Suitable starting strategy | Main decision to document |
|---|---|---|
| Genome-wide context and transposable-element methylation | WGBS or a validated enzymatic methyl-seq workflow | Coverage across CG, CHG, CHH and repetitive regions |
| Large cohort or limited budget | Reduced-representation or targeted methylation design | Which loci and contexts are omitted by enrichment |
| Low-input or damaged material | Enzymatic conversion or targeted assay | DNA integrity, conversion performance, and comparability to reference data |
| Candidate-locus validation | Targeted bisulfite or enzymatic methylation sequencing | Primer/capture design, sequence polymorphisms, and locus-level coverage |
Method selection should not be based on “single-base resolution” alone. Ask whether the assay captures the sequence contexts and genomic elements that support the biological claim. For a targeted follow-up, the Targeted DNA methylation analysis service and Target Bisulfite Sequencing resources can be considered after candidate loci are defined.
QC and Methylation Calling: Context Before Conclusions
The first analysis layer should describe whether the libraries are technically interpretable. Review conversion efficiency, read quality, adapter contamination, mapping rate, duplicate behavior, coverage distribution, and the proportion of cytosines covered at the minimum depth required by the project. For plant genomes, also inspect how repetitive sequences, organellar reads, homopolymer-like regions, and cultivar-specific variants affect mapping.
Methylation calling should preserve context. Generate context-specific coverage and methylation summaries, then examine the distribution of methylation levels in CG, CHG, and CHH sites separately. A high global methylation value driven by CG sites does not demonstrate that CHH methylation is adequately measured. Conversely, a large number of CHH calls at very low coverage may reflect sampling noise rather than widespread regulatory change.
The reference assembly and annotation version must be recorded with the analysis. If reads from divergent cultivars are aligned to a single reference without considering polymorphism, false methylation calls or apparent DMRs can occur near sequence differences. The same principle applies when assigning DMRs to promoters, genes, or transposable elements: use an explicit priority rule when a region overlaps multiple annotations.
Figure 2: QC should show whether each methylation context is sufficiently covered and technically comparable across samples.
DMCs, DMRs, and the Difference Between a Signal and a Candidate
A differentially methylated cytosine (DMC) is a single-site result. A differentially methylated region (DMR) aggregates coordinated changes across a defined genomic interval. Both can be useful, but they answer different questions and have different sensitivity to coverage, smoothing, window size, and local polymorphism.
Use a transparent DMR workflow:
- Define minimum site coverage, missingness, effect-size, and statistical thresholds before reviewing biological labels.
- Analyze CG, CHG, and CHH contexts separately or use a model that explicitly accounts for context.
- Control multiple testing and report both statistical significance and methylation difference.
- Check whether a DMR is supported by multiple cytosines and multiple biological replicates.
- Annotate DMRs against promoters, gene bodies, intergenic regions, and transposable elements using the same genome build.
Do not promote every statistically significant DMC into a mechanistic candidate. A candidate DMR should have adequate coverage, reproducible direction, a defined genomic context, a plausible relationship to the stress or trait, and an orthogonal validation plan.
Connect DMRs to Genes, Transposable Elements, and Expression
Gene assignment is a hypothesis-building step, not proof of regulation. Nearest-gene annotation can be misleading when a DMR lies in a distal regulatory region or a transposable element. Report the region class, distance to gene features, overlap with known regulatory elements, and whether the relevant gene is expressed in the sampled tissue.
| Integration layer | What it can show | What it cannot prove alone |
|---|---|---|
| DMR annotation | Which promoters, gene bodies, intergenic regions, or transposable elements are affected | That the nearest gene is the functional target |
| RNA-seq integration | Whether methylation changes track differential expression in the same samples | Causality or direction of regulation |
| Stress-pathway enrichment | Whether DMR-associated genes cluster in response, hormone, transport, or developmental pathways | That every enriched pathway is biologically active in the sampled tissue |
| Allele or cultivar comparison | Whether methylation and expression patterns differ with genotype | Whether sequence polymorphism or methylation is the initiating factor |
| Targeted validation | Whether selected loci reproduce the direction and magnitude of change | That a validated locus generalizes to all tissues or stresses |
When RNA-seq is available from matched samples, use it to prioritize DMR-associated genes with consistent expression evidence. The site’s Integrating RNA-seq and Epigenomic Data Analysis resource can support this layer. A plant methylation project should still report methylation and expression as separate measurements before integrating them into a candidate model.
Interpreting Stress-Response Methylation Without Over-Claiming
Stress-response methylation is often tissue-, genotype-, time-, and context-dependent. A drought-associated DMR in a leaf at 12 hours may not persist after recovery or appear in roots. A cultivar-specific DMR may reflect genetic variation, epiallele structure, or different developmental responses rather than a universal drought mechanism.
Published plant studies illustrate why the analysis needs this context. Rice desiccation and salinity work has linked CG methylation in gene bodies and CHH methylation in distal promoters with expression patterns, while many DMR–DEG relationships were cultivar-specific. Drought studies in mulberry and more recent work in rapeseed likewise show that the distribution of changes across CG, CHG, and CHH can vary by species, genotype, and stress design.
The defensible conclusion is usually that a methylation pattern is associated with a defined stress response in a defined biological context. A causal or heritable claim requires additional evidence, such as targeted methylation validation, genetic or epigenetic perturbation, time-course consistency, and replication in an independent population or experiment.
A Validation Ladder for Plant Methylation Candidates
Move from broad discovery to focused validation in stages:
- Discovery: identify context-specific DMCs and DMRs with adequate coverage and replicate support.
- Prioritization: combine methylation effect size, genomic annotation, expression evidence, stress relevance, and sequence quality.
- Targeted confirmation: validate the methylation direction at selected loci using targeted bisulfite or enzymatic methylation sequencing.
- Functional follow-up: test expression, chromatin state, transposable-element activity, or phenotype under matched conditions.
- Replication: evaluate the candidate in another biological replicate set, tissue, time point, or genotype before generalizing.
Figure 3: A validation ladder keeps discovery-scale methylation signals separate from locus-specific functional claims.
Common Analysis Failures
- Pooling CG, CHG, and CHH too early: combined summaries can conceal a context-specific mechanism or a coverage problem.
- Calling DMRs from one sample per group: without biological replication, significance and effect size are difficult to interpret.
- Ignoring the reference and SNP context: cultivar variation can affect alignment, methylation calling, and DMR boundaries.
- Assigning every DMR to the nearest gene: distal regulatory regions and transposable elements may be more relevant than proximity.
- Treating methylation-expression correlation as causation: stress can change both readouts through a third process, including cell composition or developmental state.
- Overlooking technical comparability: conversion efficiency, sequencing depth, and coverage differences can produce apparent context-specific changes.
What a Complete Plant Methylation Analysis Should Deliver
A decision-ready report should include the sample sheet, genome and annotation versions, conversion and mapping QC, context-specific methylation summaries, DMC and DMR tables, effect sizes and statistical thresholds, annotation tracks, pathway results, and an integrated candidate table. For each candidate, record the context, region class, methylation direction, expression direction if available, replicate support, stress condition, and proposed validation assay.
For studies intended to compare cultivars or stress time points, include a design matrix that makes genotype, treatment, tissue, and time explicit. This is more useful than a single heatmap because it lets reviewers and collaborators distinguish a robust interaction from a pattern driven by one comparison.
What This Adds to Plant DNA Methylation Studies
Plant methylation analysis becomes more informative when CG, CHG, and CHH are treated as separate evidence layers rather than merged into one percentage. Researchers planning a plant stress, development, or cultivar-comparison project can review the genome-wide DNA methylation analysis service, Epigenomic Data Analysis, and targeted validation options to align assay choice, context-aware DMR analysis, and follow-up testing. Services are provided for research use only.
FAQ
1) What do CG, CHG, and CHH mean in plant methylation analysis?
CG refers to cytosine followed by guanine. CHG and CHH are non-CG contexts in which H is a base other than guanine. They should generally be summarized and analyzed separately because their distribution, maintenance, coverage, and biological interpretation differ.
2) Is WGBS required for every plant methylation study?
No. WGBS is useful when broad genome-wide and context-specific coverage is needed, but reduced-representation, enzymatic, or targeted assays may be more appropriate for limited budgets, low-input material, or a focused candidate question. The method should match the regions and contexts needed for the claim.
3) What is the difference between a DMC and a DMR?
A DMC is a single differentially methylated cytosine, while a DMR is a region-level pattern built from multiple sites or a defined genomic window. DMRs can provide stronger regional context, but both require adequate coverage, biological replication, and transparent statistical thresholds.
4) Can a methylation change prove that a gene responds to stress?
No. Methylation and expression can be associated without one causing the other. Stronger interpretation comes from matched methylation and RNA-seq data, time-course or genotype evidence, targeted methylation validation, and functional testing under controlled conditions.
5) Why should transposable elements be included in plant methylation analysis?
Transposable elements are major methylated regions in many plant genomes and can influence genome stability and nearby gene regulation. Excluding them may remove biologically relevant stress-associated methylation changes, but annotation and mapping quality should be checked carefully because repetitive sequence can complicate analysis.
References
- Lister R, O’Malley RC, Tonti-Filippini J, et al. Highly integrated single-base resolution maps of the epigenome in Arabidopsis. Nature. 2008. doi:10.1038/nature06705.
- Cokus SJ, Feng S, Zhang X, et al. Shotgun bisulphite sequencing of the Arabidopsis genome reveals DNA methylation patterning. Nature. 2008. doi:10.1038/nature06759.
- Zemach A, McDaniel IE, Silva P, Zilberman D. Genome-wide evolutionary analysis of eukaryotic DNA methylation. Science. 2010;328(5980):916–919. doi:10.1126/science.1186366.
- Rajkumar MS, Shankar R, Garg R, Jain M. Bisulphite sequencing reveals dynamic DNA methylation under desiccation and salinity stresses in rice cultivars. Genomics. 2020;112(5):3537–3548. doi:10.1016/j.ygeno.2020.04.005.
- Whole-genome bisulfite sequencing methylome analysis of mulberry reveals epigenome modifications in response to drought stress. PubMed record. 2020.
- Integrated epigenomic and transcriptional profiling reveals genotype-specific adaptive reprogramming to drought stress in Brassica napus. PubMed record. 2025.
Research Use Only Statement
The information provided in this article is for research use only and is not intended for use in diagnostic or therapeutic procedures. CD Genomics provides sequencing and bioinformatics services for research purposes. Researchers should consult the appropriate regulatory guidelines for their specific applications.
