Short-Read vs Long-Read mtDNA Sequencing: Which Is Better for Forensic Casework?

A 2026 pedigree-based study comparing short- and long-read mitochondrial genome sequencing found no single winner: long reads improved coverage uniformity and NUMT discrimination, while short reads performed better for low-level heteroplasmy.

11 Min Read
#image_title

Short-read sequencing or long-read sequencing: which is better for forensic mitochondrial DNA analysis? A new pedigree-based study suggests that the answer is not one platform or the other. Each technology solves a different forensic problem.

Published in the International Journal of Legal Medicine on October 2, 2026, the study compared Illumina short-read sequencing with QitanTech Nanopore long-read sequencing using 177 individuals from 39 maternal pedigrees. The researchers evaluated whole-mitochondrial-genome data with particular attention to coverage, nuclear mitochondrial DNA segments (NUMTs), consensus haplotypes and heteroplasmy.

The main finding is practical: long-read sequencing performed better for coverage uniformity and distinguishing NUMT interference, while short-read sequencing showed higher accuracy for low-level heteroplasmy detection. The authors therefore proposed a complementary workflow rather than treating the technologies as direct replacements for one another.

Why this comparison matters in forensic mtDNA analysis

Mitochondrial DNA remains important in forensic genetics because each cell contains many more copies of mtDNA than nuclear DNA. That makes mtDNA particularly valuable when samples are limited, highly degraded, old or derived from hair shafts, bones and teeth.

But mitochondrial genome sequencing presents analytical challenges that differ from routine autosomal STR profiling. Two of the most important are heteroplasmy and NUMT interference.

Heteroplasmy occurs when more than one mtDNA sequence variant is present within an individual. At low levels, distinguishing genuine heteroplasmy from sequencing noise becomes difficult. NUMTs, on the other hand, are mitochondrial-like sequences that have become incorporated into the nuclear genome. Because they resemble mtDNA, they can interfere with alignment and variant interpretation.

These are precisely the areas where short and long reads behave differently.

For a broader introduction to the marker system, see our guides to mtDNA in forensic science and mtDNA versus nuclear DNA.

Study design: 177 individuals across 39 maternal pedigrees

Chen and colleagues compared two sequencing strategies using samples from 177 individuals representing 39 maternal pedigrees.

The short-read arm used an Illumina sequencing platform. The long-read arm used QitanTech Nanopore sequencing with a single-amplicon mitochondrial genome enrichment strategy.

The use of maternal pedigrees is important. Because mtDNA is maternally inherited, relatives within a pedigree provide a biological framework for evaluating whether observed sequence differences are plausible inherited variants or more likely analytical noise.

The researchers compared the platforms in several areas relevant to forensic casework:

  • coverage uniformity across the mitochondrial genome;
  • off-target sequence alignment;
  • interference from NUMTs;
  • concordance of consensus mitochondrial haplotypes;
  • accuracy of low-level heteroplasmy detection.

Long reads produced more uniform mitochondrial genome coverage

One of the strongest advantages of the long-read workflow was coverage uniformity.

The study reported a median coverage uniformity of 100% for the long-read approach. The authors attributed this in part to the use of a single-amplicon enrichment strategy, which avoided some of the uneven coverage associated with multiplex PCR used in the short-read workflow.

Coverage uniformity matters because poorly covered regions can complicate variant calling, reduce confidence in sequence interpretation or require additional analytical review.

For forensic laboratories working with challenging samples, consistent coverage across the mitochondrial genome can therefore be a meaningful operational advantage.

Long reads were better at separating mtDNA from NUMT interference

The second major advantage of long-read sequencing was its ability to distinguish genuine mitochondrial sequence from nuclear mitochondrial DNA segments.

NUMTs can be difficult because portions of the nuclear genome share substantial sequence similarity with mtDNA. Short reads originating from these regions may align incorrectly to the mitochondrial reference sequence.

The study found pronounced off-target alignment in the short-read data, with interference notably enriched on chromosomes 17 and 2. Longer reads provided more surrounding sequence context, making it easier to distinguish reads originating from the mitochondrial genome from homologous nuclear regions.

What this means for forensic interpretation is straightforward: read length can provide contextual information that reduces ambiguity when mtDNA and NUMT sequences resemble one another.

But short reads were better for low-level heteroplasmy

This is where the result changes direction.

Although long-read sequencing performed well for structural characterization and NUMT discrimination, the short-read platform showed higher accuracy for detecting low-level heteroplasmy.

The pedigree-based comparison revealed that the current long-read workflow generated background-noise signals that could be misclassified as heteroplasmy when a 10% allele-frequency threshold was applied.

That finding is important because low-level heteroplasmy can affect forensic comparison and interpretation. A minor sequence variant should not be reported as genuine simply because it crosses a numerical threshold if the platform’s error characteristics can produce similar signals.

Short-read sequencing currently has an advantage here because of its high base-level accuracy and mature analytical workflows for detecting low-frequency sequence variants.

Consensus haplotypes were concordant across both platforms

Despite these platform-specific differences, the study found that both sequencing approaches produced concordant consensus haplotypes.

That is reassuring. The disagreement was not about the major mitochondrial sequence profile itself. Instead, the important differences appeared in more technically difficult areas: off-target alignment, NUMT discrimination and low-level heteroplasmy.

This distinction matters in forensic practice. A platform can perform very well for consensus sequence determination while still requiring caution for minor-variant interpretation.

Short-read vs long-read mtDNA sequencing at a glance

Forensic considerationShort-read sequencingLong-read sequencing in this study
Consensus mtDNA haplotypeConcordantConcordant
Coverage uniformityMore affected by multiplex-PCR variabilitySuperior; median uniformity reported as 100%
NUMT discriminationMore vulnerable to off-target alignmentBetter contextual discrimination using longer reads
Low-level heteroplasmyHigher accuracy in this comparisonMore affected by background-noise misclassification
Structural/contextual informationLimited by shorter fragmentsGreater sequence context
Best role suggested by the authorsHigh-fidelity heteroplasmy quantificationStructural characterization and NUMT exclusion

So which technology is better for forensic casework?

The study does not support declaring one technology universally superior.

If the analytical problem is NUMT interference, sequence context or coverage uniformity, the long-read approach offers clear advantages.

If the problem is accurate measurement of low-level heteroplasmy, the short-read workflow performed better in this study.

The authors therefore proposed an integrated strategy: use long-read sequencing for structural characterization and NUMT exclusion, while relying on short-read sequencing for high-fidelity heteroplasmy assessment.

For difficult casework, especially complex or low-input samples, that complementary approach may be more defensible than assuming one sequencing technology must replace the other.

Why heteroplasmy thresholds cannot be transferred blindly between platforms

One of the most useful lessons from this paper is broader than mtDNA.

A numerical threshold is only meaningful in the context of the analytical system that produced the data.

If a laboratory validates a 10% heteroplasmy threshold on one sequencing platform, that threshold should not automatically be transferred to another platform with different error profiles, chemistry, enrichment methods and bioinformatic processing.

The 2026 EDNAP collaborative study on mtDNA heteroplasmy detection also highlighted the importance of platform and laboratory-specific reporting thresholds. Across Sanger, Ion Torrent and Illumina workflows, low-level point heteroplasmies could be present in raw data but not reported when they fell below laboratory-defined thresholds.

Together, these findings reinforce a core validation principle: heteroplasmy detection limits and interpretation thresholds must be established empirically for the complete analytical workflow.

What forensic laboratories should take from this study

For laboratories considering long-read mitochondrial sequencing, the paper gives several practical points to evaluate during validation:

  • coverage uniformity across the complete mitochondrial genome;
  • NUMT-related alignment behaviour;
  • platform-specific insertion, deletion and substitution error patterns;
  • false-positive heteroplasmy rates at different minor-allele thresholds;
  • repeatability and reproducibility of low-level variant calls;
  • concordance with an established sequencing method;
  • performance with degraded and low-template forensic samples;
  • bioinformatic filtering and review criteria.

Long-read sequencing should therefore be viewed as a technology with specific strengths rather than as an automatic upgrade over short-read sequencing.

The larger shift: complementary sequencing rather than replacement

Forensic genomics is increasingly moving away from simple platform-versus-platform comparisons. Different sequencing technologies generate different kinds of information and different error structures.

This study illustrates that clearly.

Long reads provide context. They can span homologous regions, improve structural interpretation and reduce ambiguity from NUMTs. Short reads provide highly accurate base-level measurements that remain valuable when the forensic question involves low-frequency variants.

The more useful question is therefore not Which platform wins?

It is Which platform is best suited to the specific analytical question?

For forensic mtDNA casework, the current evidence suggests that the strongest answer may be to use the two approaches strategically rather than treating them as competitors.

References

Chen A, Yang Q, Cao Y, Liu Y, Yang F, Zhang S, Li C. Pedigree-Based comparison of short-read and long-read sequencing for forensic mitochondrial genome analysis. International Journal of Legal Medicine. Published online October 2, 2026. DOI: 10.1007/s00414-026-04031-5. PubMed PMID: 42825929.

Kiesler K, Romsos E, Vallone P, Parson W. The EDNAP collaborative study on mtDNA heteroplasmy detection using Sanger and Massively Parallel Sequencing technologies. NIST publication, July 23, 2026. NIST publication page.

Share This Article
Follow:
Forensic Analyst by Profession. With Simplyforensic.com striving to provide a one-stop-all-in-one platform with accessible, reliable, and media-rich content related to forensic science. Education background in B.Sc.Biotechnology and Master of Science in forensic science.
Leave a Comment