DNA methylation is measured by exploiting a chemical difference between methylated and unmethylated cytosines in your DNA. The most widely used approach, bisulfite sequencing, is considered the gold standard. But the best method for a given project depends on whether you need a genome-wide picture, a measurement at specific genes, or a single global number representing overall methylation levels. Here’s how each major technique works and when to use it.
Bisulfite Conversion: The Core Chemistry
Most methylation measurement methods start with the same chemical trick: treating DNA with sodium bisulfite. This reagent selectively removes the amino group from unmethylated cytosine bases, converting them into uracil (which sequencing instruments read as thymine). Methylated cytosines resist this reaction and stay as cytosines. The conversion happens in three steps: bisulfite first attaches to the cytosine ring, then the amino group is stripped off by hydrolysis, and finally the bisulfite detaches, leaving uracil behind.
Methylated cytosines (specifically 5-methylcytosine) can also be converted by bisulfite, but the reaction is dramatically slower. This speed difference is what makes the technique work. After conversion, you sequence the DNA and compare it to the original reference. Every position that still reads as a cytosine was methylated; every position that now reads as a thymine was unmethylated.
Whole-Genome Bisulfite Sequencing
Whole-genome bisulfite sequencing (WGBS) applies bisulfite conversion to the entire genome and then sequences everything. It covers 85% to 90% of all CpG sites in the human genome, making it the most comprehensive method available. If your goal is a complete, unbiased map of methylation across every chromosome, this is the approach to use.
The tradeoff is cost. WGBS requires deep sequencing to produce reliable data at each individual CpG site, and for large sample sets (dozens or hundreds of individuals), the expense adds up quickly. It’s best suited for discovery projects where you don’t yet know which regions of the genome matter, or for generating a definitive reference methylation map of a particular cell type.
Reduced Representation Bisulfite Sequencing
Reduced representation bisulfite sequencing (RRBS) takes a shortcut. Before bisulfite treatment, the DNA is cut with restriction enzymes that target CpG-rich regions, and then only fragments of a certain size are kept for sequencing. This focuses your sequencing budget on the parts of the genome most likely to contain meaningful methylation signals, like gene promoters and CpG islands.
The coverage is far narrower than WGBS. In one comparison, RRBS covered roughly 4% to 12% of the genome depending on the sample type and sequencing platform, capturing around 5 to 6 million CpG sites at any depth. Of those, about 2.5 to 3 million had enough coverage (10 reads or more) for confident methylation calls. That’s a fraction of what WGBS provides, but RRBS is significantly cheaper per sample. It’s widely used in large cohort studies where hundreds of samples need to be profiled at a reasonable cost.
Methylation Arrays
Microarray-based platforms offer another cost-effective option for profiling methylation across many samples. The latest Illumina MethylationEPIC v2.0 array targets roughly 935,000 CpG sites, including over 200,000 new sites in enhancers and open-chromatin regions compared to the previous version. You won’t get the genome-wide reach of WGBS, but you’ll get highly reproducible measurements at nearly a million predefined locations.
Arrays require bisulfite-converted DNA as input. The minimum DNA needed for the conversion step is 250 nanograms, though Illumina recommends 500 to 1,000 nanograms for optimal reproducibility. Automated workflows typically need 1,000 nanograms. Formalin-fixed paraffin-embedded tissue samples (the kind stored after biopsies) can work with as little as 250 nanograms, which makes this platform practical for clinical and archival samples.
Because arrays produce standardized, quantitative data at fixed positions, they’re a popular choice for epigenome-wide association studies, where the same sites are measured consistently across thousands of people.
Methylation-Specific PCR
When you already know which gene or region you care about, you don’t need to sequence or array the whole genome. Methylation-specific PCR (MSP) uses bisulfite-converted DNA as a template and designs two sets of PCR primers: one that matches the methylated sequence (cytosines still present) and one that matches the unmethylated sequence (cytosines converted to thymines). Whichever primer set amplifies tells you the methylation status of that region.
This is a targeted, low-cost method well suited for clinical or diagnostic applications, like checking whether a tumor suppressor gene’s promoter is silenced by methylation. Quantitative versions of this technique can estimate the proportion of methylated versus unmethylated copies in a sample.
Global Methylation Measurement
Sometimes you don’t need to know where methylation occurs, just how much total methylation is present across the genome. Several non-bisulfite methods handle this.
Liquid chromatography with tandem mass spectrometry (LC-MS/MS) breaks DNA down into its individual building blocks and directly measures the ratio of methylated cytosine to total cytosine. It provides an absolute, quantitative number independent of DNA quality and can process over 500 samples in four days. ELISA-based kits offer a simpler bench-top alternative: antibodies bind to methylated cytosines, and a colorimetric readout gives a relative methylation level. These approaches are useful for screening large numbers of samples for broad changes in methylation, such as those associated with aging or environmental exposures, but they can’t tell you anything about specific genes or genomic regions.
Direct Detection Without Bisulfite
Bisulfite conversion is effective but harsh. It degrades DNA and can introduce errors. Newer long-read sequencing technologies skip the chemistry entirely. Oxford Nanopore sequencing threads a single DNA strand through a tiny protein pore and measures changes in electrical current as each base passes through. Methylated cytosines alter the current differently than unmethylated ones, allowing the instrument to call methylation directly from the raw signal on native, unconverted DNA.
This approach has two major advantages. First, it preserves long DNA molecules, which helps map methylation in repetitive regions and across large structural features that short-read methods struggle with. Second, it can potentially detect multiple types of modifications in a single run. The main limitation is that base-calling accuracy for modifications has historically lagged behind bisulfite methods, though newer pore designs and computational models are closing that gap.
Telling Apart Different Types of Methylation
Standard bisulfite sequencing has a blind spot: it cannot distinguish between 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC). Both resist bisulfite conversion and read as cytosines. This matters because the two marks have different biological roles.
Oxidative bisulfite sequencing (oxBS-seq) solves this by adding a chemical oxidation step before bisulfite treatment. The oxidant converts 5hmC into 5-formylcytosine, which then reacts with bisulfite and gets read as thymine. After this treatment, the only base that still reads as cytosine is true 5mC. By running both standard bisulfite sequencing and oxBS-seq on the same sample, you can calculate 5hmC levels at each position by subtraction.
Data Analysis After Sequencing
Generating raw sequencing data is only half the job. Bisulfite-treated reads need specialized alignment software because the conversion of cytosines to thymines makes standard aligners unreliable. Two main strategies exist. Three-letter aligners like Bismark and BS-Seeker2 convert all cytosines to thymines in both the reads and the reference genome before alignment, then reconstruct methylation status afterward. These tools tend to be more accurate. Wild-card aligners like BSMAP replace cytosines in the reference with a character that matches both C and T, which can yield slightly higher coverage but at the cost of some precision.
Once reads are aligned and methylation levels are calculated at each CpG site, the next step is usually identifying differentially methylated regions (DMRs), stretches of DNA where methylation differs between your groups of interest. Tools like MethylC-analyzer compare average methylation levels between groups using statistical thresholds, while HOME uses a different algorithmic approach. Downstream analysis can then map those DMRs to specific genomic features like promoters, exons, or transcription factor binding sites using packages such as methylKit or MethGO.
Choosing the Right Method
The choice comes down to three questions: how much of the genome do you need to cover, how many samples do you have, and what’s your budget?
- Full genome, few samples: WGBS gives the most complete picture but costs the most per sample.
- CpG-rich regions, many samples: RRBS balances coverage and cost for large cohorts.
- Predefined sites, large cohorts: Methylation arrays provide highly reproducible data at nearly a million sites with straightforward analysis.
- Specific genes or regions: Methylation-specific PCR is fast, cheap, and targeted.
- Overall methylation level: LC-MS/MS or ELISA gives a single global number without any sequencing.
- Long-range context or multiple modifications: Nanopore sequencing reads methylation directly on intact, long DNA molecules.
For most researchers entering the field, methylation arrays or RRBS offer the best starting point: manageable costs, established analysis pipelines, and large existing datasets for comparison. WGBS remains the definitive tool when completeness matters more than throughput.