Linear Genome Reference Bias Generates Systematic False-Positive Variant Classifications in Preventive Screening

Article Impact Level: HIGH
Data Quality: STRONG
Summary of  GeroScience. https://doi.org/10.1007/s11357-026-02436-z 
Dr. Gyula Richárd Nagy et al.

Points

Semmelweis University researchers conducted an observational case series evaluating whole genome sequencing data from twenty healthy adult participants using the standard linear reference GRCh38.
Automated bioinformatic processing systematically misclassified functional major alleles as false positive high impact variants in one hundred percent of participants across three distinct loci.
Computational artifacts occurred because the linear reference relies on rare reference minor alleles that mathematically misinterpret normal population dominant wild type states as genetic abnormalities.
Cross referencing with gnomAD confirmed these misclassified findings as benign alleles while warning that reliance on manual curation creates an unsustainable bottleneck for population scale screening.
Authors concluded that transitioning from linear reference assemblies to graph based pangenome references is essential to eliminate inherent alignment bias and secure accurate preventive genomics.

Summary

This study evaluated reference bias and diagnostic vulnerabilities associated with linear reference genomes in population-scale preventive genomic screening. Published in GeroScience by researchers from Semmelweis University, the observational case series investigated whether reference minor alleles (RMAs) in the standard linear reference GRCh38 systematically generate false-positive “high-impact” variant annotations during automated bioinformatic processing. Given that population-wide whole-genome sequencing is increasingly utilized to identify pre-symptomatic disease risk, the research sought to identify bioinformatic bottlenecks that could threaten the scalability and accuracy of personalized preventive medicine.

The study analyzed whole-genome sequencing data from an observational case series of 20 healthy adult participants. Automated alignment to the GRCh38 linear reference—which derives 70% of its sequence from a single individual—systematically misclassified functional major alleles as false-positive high-impact variants in all 20 participants (100% of cases) across three distinct loci (SLC37A4, CIMIP2A, and GPR33). These computational artifacts occurred because automated software mathematically defined healthy, wild-type alleles as deviations from rare reference minor alleles, a finding corroborated by cross-referencing population frequencies in gnomAD. While downstream manual curation and population-frequency filters currently manage these false-positive artifacts, this retrospective patching creates an unsustainable bottleneck for population-scale screening and introduces theoretical risks of false-negative masking.

The authors conclude that automated bioinformatic pipelines utilizing linear reference genomes systematically misclassify benign population-dominant alleles as high-impact functional variants. To eliminate these inherent diagnostic vulnerabilities at the alignment level, the authors advocate for a transition from linear assemblies to graph-based pangenome references that comprehensively capture human genetic diversity. Resolving these computational and infrastructure challenges is essential to ensure safety, accuracy, and scalability in future genomic aging research and personalized preventive medicine.

Link to the article: https://link.springer.com/article/10.1007/s11357-026-02436-z

References

Nagy, G. R., Munkácsy, G., Murmu, A., Longo, G., López, L. I., Cirigliano, V., & Győrffy, B. (2026). Beyond the linear genome: How reference bias threatens preventive medicine and geroscience. GeroScience. https://doi.org/10.1007/s11357-026-02436-z