Illumina 5-base produces high quality libraries and methylation performance when using enzymatic fragmentation as an alternative to mechanical shearing.
Foreward
The work presented in this blog post was led by Dr. Christian Pflueger, a genome and epigenome engineering researcher with more than a decade of experience developing advanced molecular and functional genomics workflows. Based in the Ryan Lister Laboratory at the Harry Perkins Institute and The University of Western Australia, Dr. Pflueger’s work focuses on DNA methylation and epigenetic engineering. Data shown here represents a subset of experiments performed as part of evaluating the Illumina 5-Base DNA Prep in collaboration with the local Field Applications Scientist team.
Introduction
The Illumina 5-Base DNA Prep workflow enables genome-wide DNA methylation profiling while retaining genomic sequence information from the same library. This creates an efficient route from genomic DNA to sequencing and downstream analysis, avoiding the severe sequence-complexity loss associated with traditional bisulfite conversion.1,2
A critical early step is fragmentation of genomic DNA. DNA is typically fragmented by Covaris mechanical shearing and purified before entry into library preparation (as stated in the product documentation). While mechanical shearing is highly reproducible, it requires access to dedicated instrumentation and adds handling time. These requirements may become limiting when scaling sample numbers or deploying the protocol in laboratories without routine access to a Covaris instrument.3
Enzymatic fragmentation offers a practical alternative. Enzymatic fragmentation methods fragment genomic DNA enzymatically and can be incorporated into standard molecular workflows without a dedicated shearing instrument. NEBNext Ultrashear® is methylation compatible and thus can fragment DNA without being blocked by methyl groups present in DNA. The central question for this study was whether NEBNext UltraShear-fragmented DNA could be transferred directly into Illumina 5-Base DNA Prep—without an intermediate cleanup or size-selection step—while maintaining final library quality and methylation performance.4,5
We compared libraries prepared from the same sample set following either Covaris mechanical shearing or NEBNext UltraShear enzymatic fragmentation. Final libraries were sequenced on a NovaSeq™ X, and downstream analysis was performed using DRAGEN™ Germline on BioInsight Platform Core (formerly Illumina Connected Analytics). The present evaluation focused on library and methylation performance; germline variant-calling sensitivity and concordance were not assessed.
The full laboratory protocol (including recommended modifications) is available on Protocols.io: Direct Enzymatic Fragmentation of Genomic DNA with NEBNext UltraShear for Illumina 5-Base DNA Prep.6
Methods summary
| Parameter | Description |
|---|---|
| DNA input | Genomic DNA with relevant controls prepared according to Illumina 5-Base DNA Prep input recommendations |
| Fragmentation methods | Covaris mechanical shearing versus NEBNext UltraShear enzymatic fragmentation |
| Post-fragmentation cleanup | Covaris: size selection before library preparation; NEBNext UltraShear: direct transfer into library preparation without size selection / cleanup |
| Library preparation | Illumina 5-Base DNA Prep |
| Optional modification | NEBNext UltraShear: Final libraries pooled and the pool is size selected via ProNex® beads |
| Sequencing platform | NovaSeq X |
| Sequencing configuration | 2 × 150 bp paired-end |
| Sequencing coverage | ≥30x (~500M reads per sample) |
| Secondary analysis | DRAGEN Germline on BioInsight Platform Core |
| Read trimming | First seven nucleotides removed from the 5′ end of both reads |
| Performance metrics | Library profiles, base composition, global methylation, methylation performance of controls, regional methylation differences, coverage metrics |
Workflow simplification
The modified workflow removes both the requirement for a mechanical shearing instrument and the post-fragmentation cleanup step. Following NEBNext UltraShear fragmentation, DNA is transferred directly into Illumina 5-Base DNA Prep and executed according to the product documentation through Index PCR and Post-PCR Clean-up. After 0.9x Post-PCR size selection with Illumina Purification beads, individual libraries can then be combined into a final pool, which can optionally be further size selected with ProNex® beads.
| Standard workflow | Modified workflow |
|---|---|
|
|
This approach reduces sample handling and avoids fragment cleanup for every sample at the beginning of the protocol. It also enables optional size selection on a single final library pool, which is advantageous for higher-throughput processing.
Results
Enzymatic fragmentation produces comparable library profiles to Covaris prepared libraries
Final library profiles generated following Covaris mechanical shearing or NEBNext UltraShear enzymatic fragmentation were compared to determine whether enzymatic fragmentation affected the final library fragments and overall profile. Both workflows generated the expected library profiles and similar yields.
Representative electropherogram traces revealed a consistent but modest difference in fragment distribution (Figure 1a, left panel). NEBNext UltraShear-derived libraries showed a broader size profile, with both shorter and longer fragments relative to the Covaris-prepared libraries (Figure 1a, right panel). This broader distribution likely reflects two features of the modified workflow: the omission of post-fragmentation cleanup and intrinsic technical differences between mechanical and enzymatic fragmentation methods.
The ProNex bead size selection enriched for a larger median library size, as shown in Figure 1b. In general, the broader profile did not have a negative impact on secondary analysis metrics, as shown in Figure 3a.
Key finding: Direct NEBNext UltraShear enzymatic fragmentation produces high-quality Illumina 5-base libraries with fragment-size distributions comparable to those generated using Covaris mechanical shearing, while showing a modestly broader insert-size profile.
Figure 1a. (Left panel) Electropherogram image of a subset of final Illumina 5-base libraries generated following Covaris mechanical shearing or NEBNext UltraShear enzymatic fragmentation. Covaris-fragmented samples underwent the standard fragment cleanup before library preparation, whereas NEBNext UltraShear-fragmented samples were transferred directly into Illumina 5-Base DNA Prep without post-fragmentation cleanup or size selection. Lane assignments: L = ladder; A1 = Sample 1, Covaris mechanical shearing; B1 = Sample 2, Covaris mechanical shearing; C1 = Sample 1, NEBNext UltraShear enzymatic fragmentation; D1 = Sample 2, NEBNext UltraShear enzymatic fragmentation. The electropherogram traces show highly comparable overall fragment distributions. Enzymatically fragmented DNA generated the expected library profile and yield. (Right panel) Comparison of representative electropherogram traces from final Illumina 5-base libraries generated using Covaris mechanical shearing or NEBNext UltraShear enzymatic fragmentation. NEBNext UltraShear-derived libraries show a broader fragment-size profile, containing both shorter and longer fragments than Covaris-generated libraries. This is consistent with the technical differences between mechanical and enzymatic fragmentation.
Figure 1b. (Left panel) Electropherogram image and comparison file of final Illumina 5-base libraries generated following optional ProNex size selection on final library pool (A1) and final pooled libraries prior to cleanup (B1). Lane assignments: L = ladder; A1 = Final library pool post clean up; B1 = Final library pool prior to clean up. (Right panel) Final libraries after cleanup show larger fragments with an average of 627 bp compared to 503 bp prior to cleanup, based on a region analysis from 200 to 1000 bp.
End-repair-associated DNA methylation bias is independent of fragmentation method
Read-level mCN/CN and mCG/CG composition was compared across Read 1 and Read 2 to determine whether enzymatic fragmentation introduced a distinct methylation bias. Both workflows (Covaris mechanical shearing and NEBNext UltraShear enzymatic fragmentation) showed highly similar profiles, including a clear composition bias within the first seven nucleotides (Figure 2a).
Trimming the first seven nucleotides substantially reduced the bias and restored a near-uniform profile across the remaining read length, especially for Read 2 (Figure 2b). Of note, the same bias was also observed in libraries prepared using NEBNext EM-seq (data not shown), indicating that it is not unique to Illumina 5-Base DNA Prep. The most likely source is end-repair and A-tailing during library construction rather than the fragmentation method or the Illumina 5-Base DNA Prep conversion chemistry.
Key finding: Removal of first seven nucleotides from both reads, but especially from Read 2, is recommended for both Covaris mechanical shearing and NEBNext UltraShear enzymatic fragmentation.
Figure 2a. Read-level mCN/CN and mCG/CG composition across Read 1 and Read 2 generated from Illumina 5-base libraries prepared following either Covaris mechanical shearing or NEBNext UltraShear enzymatic fragmentation. The sequence-composition within the first seven nucleotides of each sequencing read is highlighted by the red box.
Figure 2b. Read-level mCN/CN and mCG/CG composition following trimming of the first seven nucleotides from Read 1 and Read 2. Following trimming, composition profiles are highly uniform across the remaining read length for both fragmentation methods.
Global DNA methylation values are highly concordant
Global and regional DNA methylation values were compared to determine whether the fragmentation method introduced technical biases. Overall methylation context distributions were comparable across NEBNext UltraShear- and Covaris-generated libraries (Figure 3a). As expected, CpG methylation accounted for the majority of methylated cytosines, while CHG and CHH methylation remained low and consistent across methods. Unknown methylation calls also remained low.
Regional differential methylation analysis between two identical DNA samples, one processed with Covaris and one with NEBNext UltraShear, were subsequently subjected to differentially methylated region (DMR) calling. Two different DMR callers were used, metilene and DSS, comparing the fragmentation methods. After intersecting overlapping regions that passed statistical thresholds (qValFDR < 0.05 for metilene and absolute value of areaStat > 25 for DSS), 80 DMRs were identified between fragmentation methods. Two DMRs are shown in Figure 3b. An important caveat is that DMR calling is only based on single replicates (n=1) and therefore do not comprehensively capture variations between the two fragmentation methods.7,8
Although the results support overall concordance, samples intended for direct comparison should, where possible, be prepared using the same fragmentation method. Cross-method comparisons should be first benchmarked and interpreted with caution.
Key finding: NEBNext UltraShear enzymatic fragmentation preserves global methylation levels. The limited number of DMRs detected due to technical differences in fragmentation methods is minor relative to biologically significant DMRs.
| Sample | Methylated CpG | % CpG | Methylated CHG | % CHG | Methylated CHH | % CHH | Unknown | % Unknown |
|---|---|---|---|---|---|---|---|---|
| Sample 1 NEBNext UltraShear | 408 M | 81.29% | 69 M | 2.91% | 135 M | 1.60% | 50,908 | 1.81% |
| Sample 2 NEBNext UltraShear | 684 M | 80.06% | 118 M | 2.94% | 218 M | 1.52% | 83,632 | 1.78% |
| Sample 3 NEBNext UltraShear | 579 M | 79.13% | 98 M | 2.84% | 182 M | 1.48% | 70,886 | 1.76% |
| Sample 1 Covaris | 792 M | 77.09% | 136 M | 3.00% | 266 M | 1.69% | 94,167 | 1.72% |
| Sample 2 Covaris | 815 M | 76.51% | 145 M | 3.11% | 265 M | 1.64% | 94,396 | 1.72% |
| Sample 3 Covaris | 685 M | 75.41% | 118 M | 2.95% | 217 M | 1.55% | 80,307 | 1.59% |
Figure 3a. Methylation context summary for Illumina 5-base libraries generated using NEBNext UltraShear enzymatic fragmentation or Covaris mechanical shearing. The number (M = million) and percentage of methylated CpG, CHG and CHH sites, together with unknown methylation calls, are shown for three samples generated using each fragmentation method.
Figure 3b. Two representative differentially methylated regions identified between Covaris mechanical shearing and NEBNext UltraShear enzymatic fragmentation. Blue tracks represent Covaris-generated libraries and purple tracks represent NEBNext UltraShear-generated libraries. The DMRs are highlighted by black boxes. These examples illustrate the minor regional methylation differences associated with fragmentation method.
Performance of Methylation levels in spiked-in controls remains consistent
Methylation levels from control DNA provide an important read out of conversion efficiency of the enzymes in the Illumina 5-Base DNA Prep kit. The fully methylated pUC19 control and unmethylated lambda control showed comparable results following both fragmentation workflows. No systematic difference was observed between Covaris and NEBNext UltraShear libraries.
These findings indicate that enzymatic fragmentation does not adversely affect methylation values.
Key finding: Fully methylated and unmethylated controls perform equivalently following Covaris mechanical shearing and NEBNext UltraShear enzymatic fragmentation.
Figure 4. DNA Methylation performance in pUC19 and lambda controls following Covaris mechanical shearing and NEBNext UltraShear enzymatic fragmentation. A) observed mCpG methylation levels for the fully methylated pUC19 control. B) observed mCpG methylation levels for the unmethylated lambda control. Control performance was highly comparable between fragmentation methods, with high CpG methylation in pUC19 and low residual methylation in lambda across all libraries.
Secondary analysis metrics including sequencing coverage remain comparable
Sequencing metrics were reviewed to determine how the broader NEBNext UltraShear library profile translated in secondary analysis metrics. %Q30 scores were consistently high, mapping rates exceeded 99%, % uniquely mapped was comparable and generally >70%, and depth-matched samples showed similar percentages of genomic bases covered at ≥20×. One sample prepared with NEBNext UltraShear enzymatic fragmentation had lower overall sequencing depth and had consequently a lower percentage of bases at ≥20×. This difference was thus sample-related and not attributed to the fragmentation method used to generate the library.
Median insert size differed between methods, consistent with the broader electropherogram profile observed for NEBNext UltraShear libraries. These differences did not produce a systematic loss of sequencing quality or methylation performance. In general, insert sizes were longer for libraries generated via Covaris shearing. If longer inserts are desired for NEBNext UltraShear enzymatic fragmentation, we recommend optimizing the fragmentation time and performing the final library size selection as indicated in the protocol.6
Germline variant calling was not benchmarked, as it was not required for this application. Comparable coverage uniformity and other secondary analysis metrics suggest that similar performance may be expected, but sensitivity, precision, recall and concordance require dedicated validation using suitable reference material.
Key finding: Sequencing quality, mapping, and coverage metrics showed no systematic fragmentation-dependent reduction.
| Sample | Mean WGS Depth | % bases ≥20× | % Q30 | % mapped | % unique | Properly paired | Median insert size |
|---|---|---|---|---|---|---|---|
| Sample 1 NEBNext UltraShear | 21.15× | 58.90% | 95.82% | 99.28% | 71.85% | 96.92% | 307 bp |
| Sample 2 NEBNext UltraShear | 36.20× | 90.85% | 95.64% | 99.42% | 70.76% | 97.28% | 305 bp |
| Sample 3 NEBNext UltraShear | 30.70× | 87.35% | 95.50% | 99.27% | 70.64% | 96.93% | 320 bp |
| Sample 1 Covaris | 35.81× | 91.13% | 95.50% | 99.66% | 77.41% | 97.25% | 362 bp |
| Sample 2 Covaris | 36.06× | 92.44% | 94.95% | 99.64% | 72.71% | 97.44% | 420 bp |
| Sample 3 Covaris | 31.50× | 91.29% | 94.65% | 99.51% | 71.91% | 97.07% | 447 bp |
Figure 5. Secondary analysis metrics for Illumina 5-base libraries generated using NEBNext UltraShear enzymatic fragmentation and Covaris mechanical shearing. Sequencing was performed on NovaSeq X and downstream analysis used DRAGEN Germline on ICA. Data in Figure 5 is not from the same libraries shown in Figure 1b.
Considerations for germline variant analysis
The present evaluation focused on DNA methylation performance, library profile, yield, and related secondary analysis metrics. Germline variant-calling performance was not formally benchmarked, so direct conclusions cannot be made about variant sensitivity, precision, recall or concordance between fragmentation methods.
Conclusions
NEBNext UltraShear enzymatic fragmentation can be integrated into Illumina 5-Base DNA Prep as a practical alternative to Covaris mechanical shearing. The modified workflow removes the requirement for additional fragmentation instrumentation, while retaining the option to perform a single ProNex size selection on a pool of final libraries after the 0.9x IPB clean-up.
NEBNext UltraShear libraries exhibited a modestly broader fragment-size distribution, but this did not adversely affect sequencing quality, %CpG levels in global and regional contexts and in the spiked in controls. The methylation bias in the first seven nucleotides for Read 2 was independent of fragmentation method and is best addressed by routine trimming before methylation analysis.
Global DNA methylation levels were concordant between fragmentation methods. Only approximately 80 DMRs were detected between fragmentation workflows, indicating that fragmentation method contributed a relatively small technical bias. Nonetheless, when comparing samples within a study, the best practice recommendation is to use the same fragmentation method on all samples.
Overall, direct NEBNext UltraShear enzymatic fragmentation provides a robust and streamlined alternative for Illumina 5-Base DNA Prep, particularly for laboratories seeking to reduce workflow complexity, sample handling, and dependence on specialized mechanical shearing instrumentation.
References
- Illumina. From Sample to Insight: Streamlining Analysis with the Illumina 5-base solution. Illumina article.
- Illumina. Illumina 5-Base DNA Prep Data Sheet. Document M-GL-03689. Data sheet.
- Illumina. Illumina 5-Base DNA Prep Support Documentation. Support documentation.
- Illumina. TruSight Oncology 500 v2: Advantages of Enzymatic Fragmentation. Technical note M-GL-03606.
- New England Biolabs. NEBNext UltraShear: Fast and Flexible Enzymatic DNA Fragmentation for Methylation and FFPE Workflows. New England Biolabs product sheet.
- Pflueger C, et al. Direct Enzymatic Fragmentation of Genomic DNA with NEBNext UltraShear for Illumina 5-Base DNA Prep. Protocols.io. Direct Enzymatic Fragmentation of Genomic DNA with NEBNext UltraShear for Illumina 5-Base DNA Prep.
- Jühling F, et al. metilene: fast and sensitive calling of differentially methylated regions from bisulfite sequencing data. Genome Research journal.
- Feng H, Wu H. Differential methylation analysis for bisulfite sequencing using DSS. Quantitative Biology journal.