One Blood Test, Two Answers: Nanopore Sequencing Reads Genes and Methylation Together

A single sequencing run that reads both the genetic code and its chemical methylation marks has resolved three long-standing diagnostic puzzles in dystonia, according to a brief report published in Clinical Epigenetics. Researchers in Germany, Austria and the Czech Republic combined Oxford Nanopore Technologies long-read genome sequencing with a validated methylation-based classifier for KMT2B-related dystonia, confirming one uncertain variant, refuting a false-positive call and uncovering a pathogenic variant that standard whole-exome sequencing had completely missed.

KMT2B-related dystonia is a major cause of early-onset generalized dystonia, a movement disorder marked by sustained involuntary muscle contractions that can progressively distort posture and gait. Making a definitive molecular diagnosis, however, has remained difficult. The KMT2B gene is large, spanning 37 exons across roughly 20,876 base pairs and containing several GC-rich regions that make the locus prone to sequencing artefacts and missed calls in standard short-read pipelines. Non-truncating alleles, particularly novel missense substitutions, add a further interpretative hurdle, leaving a high proportion of variants classified as variants of uncertain significance. That uncertainty prolongs the diagnostic odyssey for families and can delay access to interventions such as deep brain stimulation, to which patients with KMT2B-related disorders often respond favourably.

The study team enrolled three unrelated individuals—one female and two males—with different forms of dystonia who remained molecularly unresolved after standard-of-care testing, including whole-exome sequencing. All were analysed in 2025 through a research program for rare monogenic dystonias at the Technical University of Munich and Helmholtz Munich, with written informed consent obtained under institutional ethics approval. The cases were deliberately chosen to represent distinct diagnostic challenges: a missense variant of uncertain significance, an ambiguous insertion-deletion call, and a clinical picture strongly suggestive of KMT2B-related disease despite negative prior genetic analyses.

The technical workflow centred on a PromethION 24 sequencer running R10.4.1 chemistry for 72 hours, with libraries prepared from 1,000 nanograms of genomic DNA. Basecalling used the super-accurate model with simultaneous calling of 5-methylcytosine and 5-hydroxymethylcytosine base modifications, and variants were called with the automated wf-human-variation pipeline incorporating Clair3. Because nanopore reads carry their methylation information and were tagged by haplotype, the team could compute methylation beta values per reference CpG site directly from the sequencing data—no separate array experiment required. Of 113 CpG sites in the previously defined KMT2B episignature, 104 mapped unambiguously to the nanopore call set, and 80 survived a minimum 10-fold read-depth filter in every sample. A support vector machine classifier originally trained on array data was retrained on nanopore data from the same 38 controls and 9 cases, and validated on an independent array cohort with all samples correctly classified.

The first case illustrates how methylation evidence can tip the balance on an uncertain variant. A 49-year-old man developed an irregular jerky action tremor of the right upper limb at age 31, initially interpreted as a functional movement disorder, before progressing to both arms and later cervical dystonia at 48, without major neurodevelopmental or cognitive impairment. Whole-exome sequencing had identified a novel heterozygous missense variant, c.7699C>T, p.(Arg2567Cys), absent from gnomAD and predicted to affect an invariant amino-acid residue in the C-terminal part of the protein. With no parental samples available, inheritance could not be established and the variant remained a VUS. Nanopore sequencing confirmed the variant at 33-fold local read-depth, and simultaneous methylation profiling returned a positive KMT2B-episignature score of 0.783, above the 0.5 classification threshold, supporting reclassification as likely pathogenic and revealing an atypical, milder, adult-onset manifestation of the disorder.

The second case shows the opposite value of long reads: clearing a false alarm. A 4-year-old boy with infantile-onset generalized dystonia, intellectual disability, developmental delay, emerging ataxic features and intermittent choreoathetosis had been flagged by exome sequencing as carrying a low-confidence KMT2B indel, c.6755_6758del, predicted to cause a frameshift. The call’s technical quality was poor—29-fold read-depth at the position but only a 10 percent variant allele fraction and a genotype quality of 40. High-fidelity nanopore data at 33-fold local depth showed no evidence of the variant on either allele, and visual inspection confirmed the original exome finding as a false positive, consistent with alignment artefacts known to arise in short-read data. The patient’s episignature score was 0, clustering strictly among controls, and his molecular diagnosis remains unresolved.

The third case demonstrates the blind-spot problem most dramatically. A 15-year-old girl with a classic KMT2B presentation—childhood-onset generalized dystonia beginning in the legs, microcephaly and intellectual disability—had shown significant motor improvement after bilateral deep brain stimulation of the internal globus pallidus at age 11, yet trio whole-exome sequencing had found no causative variant. The reason became clear with long reads: the beginning of KMT2B exon 1, a GC-rich region, is a known sequencing blind spot for several commercial exome enrichment kits, including the Twist kit used in her initial workup, and exome data showed no coverage there. Nanopore sequencing readily identified a heterozygous, recurrent pathogenic frameshift indel, c.17_23dup, p.(Ser9Argfs*109), at 35-fold local depth. Its pathogenicity was corroborated by an episignature score of 0.973, close to the maximal value.

Episignatures themselves deserve explanation. Pathogenic variants affecting KMT2B expression and function disrupt the protein’s known role in transferring methyl groups to lysine 4 of histone H3, which indirectly alters cytosine-phosphate-guanine methylation patterns across the genome. The result is a disease-specific DNA methylation fingerprint that can be used to prioritize and reclassify variants of uncertain significance. Traditionally, this analysis requires a separate, array-based methylomics test performed after genetic testing, adding cost and turnaround time. The new workflow extracts high-resolution methylation data from the very same sequencing run used for variant calling, collapsing two sequential tests into one and making what the authors call a “one-test-fits-all” approach a practical reality.

The authors are careful about scope. This was a brief report of three selected cases within a research program, not a clinical validation study; the SVM pipeline is undergoing rigorous prospective testing in a larger, independent cohort before any general clinical deployment, and the research-stage analysis pipeline is archived on Zenodo. The classifier itself was retrained and validated on previously published cohorts, and the three index cases represent unpublished findings not included in the team’s recent whole-cohort reporting. Case 2, notably, left without a molecular diagnosis, a reminder that a negative episignature and a refuted variant call resolve uncertainty but do not always end the search.

Even with those caveats, the implications for rare-disease diagnostics are considerable. In a single workflow, long-read sequencing resolved a VUS by supplying functional methylation evidence, eliminated a false-positive indel, and captured a pathogenic variant inaccessible to short-read exomes—each failure mode that typically stalls dystonia diagnostics. The authors conclude that nanopore long-read sequencing should be considered as a primary diagnostic tool for suspected KMT2B-related dystonia, particularly when standard pipelines are inconclusive, potentially shortening the diagnostic odyssey for affected families and enabling timely, precision clinical management. If prospective validation succeeds, the same logic of reading sequence and epigenotype together could extend well beyond dystonia to the broader family of episignature-defined neurodevelopmental disorders.

Subject of Research: Using nanopore long-read sequencing with a KMT2B methylation episignature classifier to diagnose dystonia

Article Title: Nanopore long-read sequencing facilitates accurate diagnosis of KMT2B-related dystonia

Article References: Sorrentino, U., Mirza-Schreiber, N., Pavlov, M., Peymani, F., Havrankova, P., Indelicato, E., Serranová, T., Kunc, L., Lösecke, S., Tilch, E., Eck, S., Prokisch, H., Winkelmann, J., Boesch, S., Jech, R., Graf, E., Oexle, K., & Zech, M. (2026). Nanopore long-read sequencing facilitates accurate diagnosis of KMT2B-related dystonia. Clinical Epigenetics, 18(1), Article 167. https://doi.org/10.1186/s13148-026-02243-5

Image Credits: AI Generated

DOI: 10.1186/s13148-026-02243-5

Keywords: KMT2B, dystonia, nanopore sequencing, long-read sequencing, DNA methylation, episignature, variant of uncertain significance, whole-exome sequencing, deep brain stimulation, genetic diagnostics, Clinical Epigenetics, Nanopore