What a clearer view of epigenetics can mean for cancer research

Cancer is not driven by DNA sequence mutations alone. Chemical modifications layered onto DNA can reshape gene regulation without altering the underlying code, silencing tumor suppressors, activating oncogenes, and rewiring cellular identity. To fully understand tumor biology, researchers need to see both the genome and its epigenetic state at once, ideally at single-molecule resolution.

Long-read sequencing chemistry and software continue to expand what can be detected from native DNA in a single assay. Today, HiFi sequencing on Revio and Vega systems reports two epigenetic marks directly from genomic DNA – 5mC and 6mA – and the SPRQ-Nx chemistry and software release planned for mid-year now adds 5hmC to the mix. This release will also offer improved 5mC performance, enabling deeper insight particularly in cancer, where methylation patterns often define tumor type and disease state.

These advances are broadening what researchers can expect from HiFi data and setting a new bar for what a single assay can reveal.

Epigenetic modifications are stable chemical changes to DNA that do not alter the underlying sequence. These changes influence how genes are regulated, helping determine which regions of the genome are active and which remain silent.

In humans, the most common DNA epigenetic modification is 5-methylcytosine or 5mC. This mark is typically associated with silenced genomic regions, while active gene promoters are often unmethylated. DNA methylation patterns are faithfully maintained as cells divide, yet they vary substantially between cell types. These cell-type-specific patterns are essential for development and differentiation, and when disrupted, can contribute to disease. Because methylation plays such a central role in regulating genome function, accurately measuring it is critical to understanding both normal biology and disease.

A common approach to measuring methylation with DNA sequencing is to convert the epigenetic mark into a change in a DNA base pair using bisulfite or enzymatic conversion. Although prevalent, this approach requires additional steps in library preparation and can induce DNA damage, limiting read length and haplotype resolution. It also fundamentally changes the sequence of the molecule being measured, disrupting the ability to measure sequence and methylation status in the same molecules. These alterations can obscure important biological patterns, particularly in cancer, where spontaneous C T deamination events can be difficult to distinguish from these protocol-driven changes.

In contrast, HiFi sequencing detects DNA methylation directly on native molecules without any additional library preparation steps or sequence alterations. As the polymerase synthesizes DNA, distinctive changes in kinetic measurements such as incorporation timing reflect the presence of modified DNA bases on the template strand. These subtle kinetic signatures can be modeled computationally to identify methylated sites while preserving long range context and haplotype structure.

Researchers have used HiFi sequencing to link genetic and epigenetic variation , identify disease causing variants , and characterize epigenetic signatures across tissues and conditions. By combining accurate sequence and methylation information in a single experiment, HiFi sequencing has enabled a more comprehensive multiomic view of genome regulation.

Notably, the upcoming SPRQ-Nx chemistry and software upgrade brings measurable improvements to 5mC detection performance. By applying updated modeling and training strategies, this release substantially increases concordance with whole genome bisulfite-based approaches when evaluated as aggregate methylation across molecules and delivers strong single-molecule accuracy in synthetic test datasets.

Accuracy in synthetic test molecules

Correlation to WGBS, HG002 *

*at 60X coverage

If 5mC methylation set the stage, 5-hydroxy methyl cytosine (5hmC) is a newer counterpart now expanding the story. Formed through oxidation of 5mC, 5hmC is present in humans and is particularly enriched in post-mitotic tissues such as brain, as well as during embryonic development. It serves as an important biomarker and is thought to play distinct regulatory roles, yet it remains less well understood and has been difficult to measure.

Enabled by new AI models based on the HK2 deep learning framework, PacBio is extending methylation detection to include 5hmC alongside 5mC later this year.

In training and test datasets, the caller accurately identifies per-strand 5hmC, maintaining high recall at a low false positive rate. This is key, as 5hmC occurs at a low rate in human DNA and a low false positive rate is required to identify a true biological signal. With the selected default parameters (rightmost purple datapoint below), the caller delivers 80% recall and a 0.86% false positive rate. Further, each call is assigned a confidence score, which can be used to increase stringency and provide an even lower false positive rate, based on application needs.

This high recall and low false positive rate allow detection of 5hmC levels in biological tissues. For example, in applying this caller to various mouse tissues in preliminary experiments , we observe higher rates of 5hmC in brain and kidney and negligible levels in thymus, as expected from bulk ELISA experiments.

These methylation detection advances have broad implications, particularly for cancer research . Tumors frequently display distinct methylation signatures, where loss of methylation can awaken oncogenes and localized hypermethylation can silence tumor suppressor genes, reshaping gene expression programs that drive disease.

Most human cell lines show approximately 70% CpG methylation. In contrast, new HiFi sequencing data in the cancer cell line K562 shows global CpG methylation closer to 35%. This striking reduction highlights the scale of epigenetic reprogramming that can occur in cancer, as evidenced in the hypomethylation of the PLAU promoter in K562 in the example below.

As another example in a sarcoma sample, HiFi sequencing observed a 29 kb deletion in haplotype 1 encompassing much of the BRCA2 locus, and hypermethylation of the BRCA2 promoter in haplotype 2. This hypermethylation led to loss of BRCA2 gene expression from the remaining allele, effectively acting as a second hit in this sample. Importantly, short-read sequencing did not reveal this causal mutation.

Methylation detection with HiFi sequencing also extends beyond cytosine. The Revio software also detects N6 methyladenine or 6mA, and updates to this caller are planned for this year. Although 6mA is not natively present in human DNA, it is widely used as a biochemical tool to measure chromatin accessibility in assays such as Fiber-seq , where 6mA marks open chromatin. Continued improvements to 6mA detection further expand the range of epigenomic applications supported by a single sequencing workflow.

Together with 5mC, 6mA, and 5hmC detection, these latest upgrades represent a shift in the expectation for how epigenetic signatures are detected in genome sequencing. HiFi sequencing now delivers accurate sequence and methylation information from a single library preparation and at the level of individual molecules.

This combination is powerful. It means researchers can phase variants, map structural changes, and profile epigenetic marks all in a single experiment, opening the door to studying the regulation and development of cancer with sharper resolution and fewer compromises.

Learn more about HiFi methylation detection at our epigenetics webpage and look out for upcoming methylation datasets .