Concordance Analysis of Minigene Assays and Patient RNA Informs Clinical Splicing Guidelines

Article Impact Level: HIGH
Data Quality: STRONG
Summary of  Genome Medicine. https://doi.org/10.1186/s13073-026-01746-3 
Dr. Daffodil M. Canson et al.

Points

University of Otago researchers co-led an international study in Genome Medicine establishing guidelines for using RNA splicing evidence to resolve uncertain genetic test results.
Evaluation of over forty one thousand genetic variants across five thousand four hundred fifty eight genes demonstrated eighty nine percent concordance between minigene models and patient derived RNA.
High throughput splicing assays exhibited significant design limitations, with only one out of seven evaluated datasets meeting performance criteria suitable for clinical diagnostic use.
Clinical calibration established that variants producing eighty percent or more abnormal RNA provided strong pathogenic evidence, whereas twenty percent or less indicated non pathogenic effects.
Authors provided a standardized framework to help clinical laboratories validate RNA assays, calibrate artificial intelligence tools like SpliceAI, and increase overall genetic diagnostic rates.

Summary

This study evaluated the clinical validity and diagnostic utility of construct-based RNA splicing assays for variant classification in clinical genomics. Published in Genome Medicine and co-led by researchers at the University of Otago alongside international collaborators, the investigation systematically calibrated laboratory minigene models against patient-derived RNA data. The research sought to establish standardized parameters for applying functional splicing evidence within international variant interpretation frameworks, addressing a critical bottleneck in resolving variants of uncertain significance.

The investigators reviewed over 41,000 genetic variants across 5,458 genes from published minigene studies. Comparisons between well-designed traditional multi-exon minigene assays and patient-derived RNA demonstrated high concordance, showing complete or strong agreement in 89% of overlapping cases. Conversely, of seven high-throughput multiplexed assays of variant effect (MPSAs) evaluated, only one met stringent performance criteria suitable for direct clinical application. Furthermore, integrating construct-based RNA data improved predictive accuracy for the artificial intelligence tool SpliceAI. Clinical calibration against ClinVar entries established that variants generating 80% abnormal RNA transcript provided strong evidence of pathogenicity, whereas variants producing 20% abnormal RNA transcript provided strong evidence against a disease-causing effect.

The authors conclude that well-designed construct-based minigene assays provide robust, quantitative functional evidence for RNA splicing disruption, particularly when patient tissue is inaccessible. However, rigorous evaluation of design limitations and performance benchmarking is required before incorporating high-throughput assay data into clinical workflows. These evidence-based recommendations establish a standardized framework for laboratories to incorporate RNA evidence into variant classification, enhancing diagnostic yield for patients with suspected inherited disorders.

Link to the article: https://link.springer.com/article/10.1186/s13073-026-01746-3

References

Canson, D. M., Wiggins, G. A. R., Velasco-Sampedro, E. A., Pearson, J. F., Walker, L. C., & Spurdle, A. B. (2026). Evidence-based recommendations for application of construct-based splicing data in clinical variant classification. Genome Medicine. https://doi.org/10.1186/s13073-026-01746-3