The ten-year evolution of Sherloc, a points-based framework for genetic variant classification.

Abstract

Variant classification underpins the clinical utility of genetic testing. Sherloc is a refinement of the 2015 ACMG/AMP guidelines that implements a structured, points-based scoring framework. It has been applied to 2.6 million variants through 19 versions (v4.2-v7.1) for 5.5 million individuals referred for genetic testing. Here, iterative updates to Sherloc over ten years were evaluated for their effect on variant classification. Refinements to the scoring framework were expected to progressively reduce the rate of variants of uncertain significance (VUS). Furthermore, it was hypothesized that points-based scores would serve as quantitative predictors of VUS reclassification outcomes. Tracking 32,241 variants initially classified with v4.2, 3,834 VUS had been reclassified by the study end; 2,801 (73%) of these relied on evidence criteria introduced after v4.2. Separately, in a simulated removal of post-v4.2 evidence criteria from 615,341 recent classifications, removing AI-related criteria altered classifications for 129,459 (21.0%) variants. The Sherloc score was strongly correlated with the log ratio of upgrades to downgrades (adjusted r2 = 0.95) and moderately correlated with the overall reclassification rate (adjusted r2 = 0.49). High-scoring VUS were disproportionately reclassified by cascade family testing or RNA analysis, although most variants reclassified by these approaches had low scores. These findings demonstrate that adaptive, points-based classification systems materially reduce VUS rates and support quantitative, score-based variant evaluation.