Reference bias in linear genome screening: 20 healthy adults' variants
In an observational case series of 20 healthy adults, automated analysis against the GRCh38 linear reference misclassified healthy major alleles as high-impact variants at three loci in all 20 participants; graph-based pangenome references are proposed as a possible fix. [1]
Automatic summary · not editorially reviewed
The key findings
An observational case series of 20 healthy adults undergoing preventive genome sequencing used automated processing against the GRCh38 linear reference to find high-impact annotations at loci where GRCh38 differs from the population consensus major allele. [1]
In all 20 participants (100%), linear alignment to GRCh38 misclassified functional major alleles as false-positive high-impact annotations at three loci: SLC37A4, CIMIP2A and GPR33. [1]
The authors describe a theoretical risk of false-negative classifications if RMAs mask true pathogenic variants, and propose graph-based pangenome references while noting computational and standardization challenges. [1]
The source at a glance
- Study type / format
- Observational case series (abstract).
- Population / context
- 20 healthy adults undergoing preventive genome sequencing.
- Summary based on
- Study abstract
- Original publication
- July 25, 2026
Screening 20 healthy adults against GRCh38
The authors describe genotype-first screening, which compares patient genomes to a standard reference. They note that the GRCh38 linear assembly contains reference minor alleles (RMAs) — variants that are rare in the population but present in the reference — and call this a structural vulnerability for automated clinical variant classification. To test whether RMAs systematically produce false-positive annotations, they conducted an observational case series of 20 healthy adults undergoing preventive genome sequencing. Automated bioinformatic processing used the GRCh38 linear reference to identify "high-impact" variant annotations at loci where GRCh38 differs from the population consensus major allele.
Supporting source [1]
Healthy alleles flagged at three loci in all 20
Among all 20 participants (100%), linear alignment to GRCh38 systematically misclassified functional major alleles as false-positive "high-impact" annotations at three distinct loci: SLC37A4, CIMIP2A and GPR33. The authors state these artifacts occurred solely because the analysis software mathematically defined the healthy wild-type state as a deviation from the rare RMA. Cross-referencing with gnomAD confirmed these as population-dominant benign alleles. The authors also describe a theoretical risk of false-negative classifications if RMAs mathematically mask true pathogenic variants; this risk is described as theoretical, not observed in this series.
Supporting source [1]
Filters, pangenome proposal and open challenges
The authors conclude that automated pipelines using linear references systematically misclassify healthy alleles as high-impact functional annotations. They note that downstream population-frequency filters manage the false-positive artifacts, but describe this retrospective patching as an unsustainable bottleneck for population-scale screening. Transitioning to graph-based pangenome references is presented as a highly promising approach to resolve these diagnostic vulnerabilities at the alignment level, though computational and standardization challenges must first be addressed to ensure accuracy in large-scale aging research. This is a proposal, not a validated or approved diagnostic product.
Supporting source [1]
How to interpret this
Abstract only; the full text was not reviewed. This is an observational case series of 20 healthy adults without a control group, so it shows misclassification in this small sample, not rates in broader or clinical populations. The false-negative risk is described as theoretical and was not observed. Graph-based pangenome references are a proposal, not a validated or approved diagnostic tool.
Based on the study abstract; the full paper, tables and supplementary data were not reviewed.
Source & supporting passages
Gyula Richárd Nagy, Gyöngyi Munkácsy, Ankita Murmu, Giuliana Longo, Luis Izquierdo López, Vincenzo Cirigliano, Balázs Győrffy · doi.org · July 25, 2026