AI Insight
Researchers have developed a new method to improve the interpretation of genetic variants by calibrating allele frequency thresholds used in clinical diagnosis. Instead of using fixed default values, they stratified variants based on inheritance mode and gene-level missense constraint, deriving evidence thresholds that vary systematically across genes. Testing on held-out genes, their stratified approach achieved 96.7% accuracy compared to 90.1% for unstratified methods, and performed comparably to expert panel-specified cutoffs while covering thousands of additional genes.
Why it matters
This approach provides empirically-derived, gene-specific thresholds for variant classification that can be applied to thousands of genes lacking expert panel guidance, potentially improving the accuracy of genetic disease diagnosis. The method offers a scalable solution to address the challenge that different genes have different tolerance levels for variation, which current one-size-fits-all approaches fail to capture.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
Allele frequency (AF) is among the most frequently applied lines of evidence in variant classification, yet the ACMG/AMP criteria that use it (BA1, BS1, PM2) are still applied at fixed defaults while computational predictors have been systematically recalibrated. Population frequencies are shaped by selection, ascertainment, and gene-level demography at once, and few genes carry enough classified variants to set a threshold directly. Extending the calibration approach applied to computational predictors, we used inheritance mode and gene-level missense constraint as stratification axes and pooled variants within each stratum. ClinVar missense variants annotated against gnomAD v4.1.1 were stratified along both, and gene-normalized kernel density estimates were fit to pathogenic and benign variants within a sliding window along the constraint axis. Thresholds were placed where the likelihood ratio crossed ACMG/AMP evidence strengths at a prior of 0.0441. Derived thresholds varied systematically with constraint and differed between inheritance modes, departing from the fixed defaults in both directions. On held-out genes, stratified cutoffs reached 96.7% accuracy against 90.1% unstratified. Restricted to the 73 ClinGen expert panel genes with autosomal dominant or recessive inheritance, the derived cutoffs reached 91.0% accuracy at 69.5% variant coverage, against 88.8% accuracy at 86.2% coverage for the panel-specified cutoffs. AF thresholds for these criteria are not constant across genes, and inheritance mode and missense constraint capture much of that variation. The resulting cutoffs are empirically derived, carry explicit uncertainty, and deploy as a lookup table across thousands of genes no expert panel currently covers.
Source: Empirically calibrated allele frequency thresholds for ACMG BA1, BS1 and PM2 evidence criteria