Summary

Standard whole-genome bisulfite sequencing (WGBS) has been the gold standard for single-base methylation analysis for over a decade. But its reliance on harsh chemical conversion — sodium bisulfite at low pH and high temperature — fragments DNA, degrades library complexity, and sets a high bar for input quantity and quality. Enzymatic methyl-seq (EM-seq) replaces bisulfite with a gentler two-step enzymatic conversion (TET2 oxidation followed by APOBEC deamination) that preserves DNA integrity. This makes EM-seq the better choice for low-input samples, degraded DNA, FFPE tissues, and circulating cell-free DNA. This article compares the two methods across the dimensions that matter most when sample material is limited. If you have a project involving challenging DNA samples, explore the EM-seq service or whole genome bisulfite sequencing service for project-specific guidance.

Key Takeaways

• EM-seq uses gentle enzymatic conversion that preserves DNA backbone integrity, producing longer inserts and higher library complexity than WGBS
• EM-seq works reliably with 1–10 ng input DNA; WGBS typically requires 100 ng–1 µg
• EM-seq provides more uniform coverage in GC-rich regions, detecting up to 32% more CpG sites at low input
• WGBS remains a cost-effective choice when ample high-quality DNA is available
• Both methods produce compatible data for standard bisulfite-seq analysis pipelines (Bismark, BWA-meth)

Figure 1. EM-seq uses enzymatic conversion (TET2 + APOBEC) that preserves DNA integrity, while WGBS relies on chemical bisulfite conversion that fragments DNA and reduces library complexity.

When Input DNA Changes the Equation

The choice between EM-seq and WGBS is not about which method produces better data from high-quality, abundant DNA. Both methods generate high-quality genome-wide methylation profiles from intact DNA at standard input levels. The decision point is reached when the sample material is limited, degraded, or both.

Three scenarios repeatedly arise in research projects:

• Low-yield samples. A laser-capture microdissection capture yields nanograms of DNA. A rare cell population sorted by flow cytometry provides 5,000–10,000 cells. An ancient DNA extract from preserved tissue contains highly fragmented, modified DNA. In each case, the input falls below the standard WGBS requirement of 100 ng–1 µg.
• Degraded DNA from FFPE. Formalin-fixed, paraffin-embedded tissues are the most common source of degraded DNA in methylation research. The fixation process crosslinks DNA with proteins and fragments it to 200–500 bp average size. Standard WGBS library preparation from FFPE DNA typically produces low-complexity libraries with high duplication rates.
• Circulating cell-free DNA. Plasma cfDNA is inherently fragmented (~166 bp) and present at low concentrations (5–50 ng per mL of plasma). Every molecule counts, and library preparation methods that further fragment or lose material are at a disadvantage.

In all three cases, the key question is not whether methylation data can be generated — both methods can produce data from low input — but which method produces higher quality data: more unique reads, higher coverage, lower duplication, and more reliable methylation calls per unit of starting material.

How Each Method Treats DNA

The fundamental difference between WGBS and EM-seq lies in how they distinguish methylated from unmethylated cytosines. One method uses chemical conversion; the other uses enzymatic conversion.

WGBS: Sodium bisulfite conversion. Treatment with sodium bisulfite at low pH (~5.0) and high temperature (~70°C) deaminates unmethylated cytosines to uracils. Methylated cytosines (5mC) are resistant to deamination and remain as cytosines. The chemical reaction is robust and well-characterized, but the conditions are harsh. The low pH and heat cause depurination and strand cleavage, fragmenting the DNA backbone. A typical WGBS library shows average insert sizes of 100–200 bp, even when starting with high-molecular-weight DNA. This fragmentation limits the effective library complexity and reduces the number of unique molecules available for sequencing.

EM-seq: Enzymatic conversion. EM-seq replaces the chemical step with two enzymatic reactions. First, TET2 oxidizes 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), and T4-BGT glucosylates 5hmC to protect it. Second, APOBEC3A deaminates unmethylated cytosines to uracils, while the protected modified cytosines remain unchanged. The entire process occurs under neutral pH and moderate temperature conditions that do not damage DNA. The result is identical to bisulfite conversion in output — unmethylated cytosines read as thymines — so the same bioinformatics pipelines work without modification. But the DNA backbone is preserved, and library insert sizes typically range from 300–500 bp.

The practical consequence of intact DNA is that EM-seq libraries contain more unique, non-duplicate molecules per nanogram of input. For samples where every picogram counts, this difference translates directly into better data.

Coverage and Data Quality at Low Input

Several benchmarking studies have quantified the performance gap between EM-seq and WGBS at reduced input amounts. The results consistently favor EM-seq when input falls below 100 ng.

Library complexity. At 10 ng input, EM-seq libraries retain approximately 60–70% unique reads after deduplication, compared with 30–40% for WGBS libraries from the same input. The difference stems directly from DNA damage during bisulfite conversion, which reduces the number of amplifiable molecules.

CpG coverage. EM-seq detects significantly more CpG sites than WGBS at equivalent sequencing depth when starting from low input. In the 1–10 ng range, the advantage is approximately 32% more CpG sites detected across all sequence contexts (CG, CHG, CHH). The gain is most pronounced in GC-rich regions, where bisulfite-induced damage disproportionately affects coverage.

Coverage uniformity. GC bias is a well-known artifact in WGBS data. Bisulfite treatment over-represents methylation levels in GC-rich regions because methylated cytosines resist conversion while unmethylated cytosines in the same regions are converted, creating sequence asymmetry. EM-seq avoids this artifact because the enzymatic conversion does not depend on GC content. Coverage across the GC spectrum is more uniform, leading to more reliable methylation calls in CpG islands and promoter regions.

Reproducibility. Technical replicate correlation for EM-seq at low input remains high (intraclass correlation coefficient >0.85), while WGBS replicate correlation drops more sharply as input decreases. This means EM-seq not only captures more data from limited material but does so more consistently.

Figure 2. EM-seq outperforms WGBS at low DNA input across key metrics including unique reads, CpG coverage, GC uniformity, and technical reproducibility.

Sample Types That Favor EM-seq

While EM-seq offers advantages across a range of challenging sample types, three categories stand out as clear candidates for choosing EM-seq over WGBS.

FFPE tissues. The combination of formalin-induced crosslinking and bisulfite-induced fragmentation makes WGBS from FFPE DNA particularly challenging. EM-seq’s gentler conversion avoids adding further damage and preserves more of the already-limited library complexity. FFPE-optimized EM-seq protocols can produce usable libraries from as little as 1–10 ng of FFPE DNA, with DV200 values as low as 20–30%. Standard WGBS typically fails below DV200 of 30%.

Circulating cell-free DNA. cfDNA’s natural fragment size (~166 bp) is close to the typical insert size of a bisulfite-converted library, meaning nearly every molecule is at risk of being lost during cleanup steps. EM-seq preserves the native fragment profile, and the longer effective insert sizes after library preparation (300–500 bp compared with cfDNA’s 166 bp) come from the adapter-ligated, full-length molecules rather than further fragmentation. EM-seq also preserves the characteristic cfDNA fragment end patterns that carry biological information about nucleosome positioning.

Precious or irreplaceable samples. Any sample that cannot be re-collected — a rare biopsy, an archaeological specimen, a neonatal dried blood spot — benefits from a method that extracts more information per molecule. EM-seq’s higher conversion efficiency at low input means more unique methylation calls per sequencing read, making it the preferred choice when sample material is the limiting factor.

Figure 3. Sample types and DNA conditions where EM-seq provides clear advantages over WGBS — FFPE, cfDNA, precious samples, and low-input DNA.

When WGBS Still Makes Sense

EM-seq is not always the better choice despite its technical advantages. WGBS remains the appropriate method in several common situations.

High-quality DNA in abundance. When input DNA is plentiful (≥500 ng) and intact (high molecular weight, no degradation), both methods produce comparable results. The difference in library complexity narrows or disappears because there is sufficient material to absorb the losses from bisulfite treatment. In this scenario, WGBS is the more cost-effective option.

Existing data comparability. If a study extends a prior WGBS dataset or needs to match data generated with bisulfite conversion, staying with WGBS avoids introducing method-related variation. While EM-seq data are compatible with WGBS analysis tools, subtle differences in coverage patterns and methylation bias could confound comparisons between datasets generated with different conversion methods.

Budget constraints per sample. WGBS reagents cost less than enzymatic conversion reagents. For large cohort studies with adequate DNA, the per-sample reagent savings with WGBS can be substantial. The trade-off is acceptable because the input requirements are met and the degradation is tolerable.

Practical Decision Framework

The choice between EM-seq and WGBS can be reduced to a few questions:

For researchers who need to maximize data yield from limited or degraded samples, EM-seq offers a clear advantage. The higher per-sample reagent cost is offset by the ability to generate usable data from samples that would fail or produce marginal results with WGBS. As enzymatic conversion reagents become more widely adopted and competition increases, the cost gap is expected to narrow.

FAQ

1) Can I use my existing WGBS bioinformatics pipeline for EM-seq data?

Yes. EM-seq produces the same output as bisulfite sequencing — unmethylated cytosines read as thymines, methylated cytosines read as cytosines. Standard tools including Bismark, BWA-meth, methylKit, and DSS work with EM-seq data without modification. The alignment parameters and methylation calling steps are identical.

2) How much DNA does EM-seq require compared to WGBS?

EM-seq typically requires 100 pg to 100 ng of input DNA, depending on the library preparation kit. Standard WGBS requires 100 ng to 1 µg. For input amounts below 100 ng, EM-seq consistently produces higher library complexity and more usable methylation data than WGBS.

3) Is EM-seq more expensive than WGBS per sample?

The enzymatic conversion reagents are more expensive than sodium bisulfite. However, the total cost per sample depends on sequencing requirements. Because EM-seq libraries have higher complexity and lower duplication rates, less sequencing is wasted on duplicate reads. For low-input samples where WGBS would require deeper sequencing to compensate for lost complexity, the total project cost may be comparable or favorable for EM-seq.

4) Does EM-seq distinguish 5-methylcytosine from 5-hydroxymethylcytosine?

No. Like bisulfite sequencing, EM-seq reads both 5mC and 5hmC as methylated. If distinguishing these modifications is necessary, an orthogonal method such as oxidative bisulfite sequencing (oxBS-seq) or chemical labeling (ACE-seq) is required. For most methylation studies where total methylation is the quantity of interest, this is not a limitation.

5) What types of degraded DNA work best with EM-seq?

EM-seq has been successfully used with FFPE DNA (DV200 as low as 20–30%), circulating cell-free DNA from plasma, DNA from laser-capture microdissection, and DNA extracted from archived or preserved specimens. The common factor is that the enzymatic conversion does not add further damage, so the library quality reflects the starting DNA quality rather than the conversion method.

Related CD Genomics Services

• Enzymatic Methyl-Seq (EM-seq) Service — Gentler methylation sequencing for low-input and degraded DNA
• Whole Genome Bisulfite Sequencing (WGBS) — Standard bisulfite-based methylation analysis
• Targeted DNA Methylation Analysis Service — Focused validation of specific regions

References

1. Olova NN, Andrews S. "Whole Genome Methylation Sequencing via Enzymatic Conversion (EM-seq): Protocol, Data Processing, and Analysis." Methods in Molecular Biology. 2025;2866:73-98. doi:10.1007/978-1-0716-4192-7_5
2. Vaisvila R, Pomaluri VKC, Sun Z, et al. "Enzymatic methyl sequencing detects DNA methylation at single-base resolution from picograms of DNA." Genome Research. 2021;31(7):1280-1289. doi:10.1101/gr.266551.120
3. Liu X, Pang Y, Shan J, et al. "Beyond the base pairs: comparative genome-wide DNA methylation profiling across sequencing technologies." Briefings in Bioinformatics. 2024;25(5):bbae440. doi:10.1093/bib/bbae440
4. Guanzon D, Ross JP, Ma C, Berry O, Liew YJ. "Comparing methylation levels assayed in GC-rich regions with current and emerging methods." BMC Genomics. 2024;25:741. doi:10.1186/s12864-024-10605-7

Services mentioned in this article are provided for research use only and are not intended for clinical diagnosis, treatment, or personal health assessment.