medRxivJClub
Transparent Implementation and Benchmarking of ACMG/AMP Variant Classification in fastVEP
medRxiv · · Preprint · Open access
Huang, Ganapathi + 1 more
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
fastVEP’s ACMG/AMP classifier, using Tavtigian points for pathogenic and the 2015 combining table for benign evidence, matched ClinVar’s five-tier classes for 431,727 variants (64.1%).
- Why it matters: ACMG/AMP is widely used, but scalable software needs operational choices that are not fully specified by the original framework or ClinGen guidance. This work targets improved agreement with expert-curated classifications without increasing opposite-direction calls.
- What they did: They implemented all 28 ACMG/AMP criteria in fastVEP and tested symmetric vs hybrid evidence-combining, class cutoffs, and population/benign thresholds on 673,660 ClinVar variants (≥ two stars) from April 2026. Clinical geneticists reviewed opposite-direction calls, then applied the selected classifier to 116,078 newly qualifying variants in the October 2026 release.
- The result: The final model produced direction agreement for 534,335 variants (79.3%) with only 51 opposite-direction calls (0.0076%); on later variants, same-direction agreement was 72.6% with 20 opposite-direction calls. This enables conservative large-scale triage that leaves variants uncertain when evidence is insufficient for expert-level resolution.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.
- Open access
Free. We’ll email you when a new paper cites this one or any other paper you follow on JClub — at most one email a day. Every email has a one-tap unsubscribe.