ACMG Classification (bioSkills)

A Claude Code skill that applies the ACMG/AMP variant-classification framework as it is actually practised today — ClinGen SVI specifications, the Tavtigian Bayesian point system, and calibrated thresholds for in-silico predictors — rather than the 2015 rules read literally.

   
Type Claude Skill
Supplier GPTomics bioSkills (community OSS, MIT)
Availability GA — part of the bioSkills collection
Pricing Free / OSS (MIT). The predictor scores and population data it reasons over (REVEL, BayesDel, AlphaMissense, SpliceAI, gnomAD, ClinVar) come from their own sources under their own terms
Capabilities Read/Write — Claude runs the skill’s Python workflow locally and queries public APIs; it is not an MCP tool
Verified works · 2026-08-17
Security caution · 2026-08-17 — GPTomics/bioSkills is now archived upstream; MIT and skill dir confirmed unchanged

How to install

bioSkills is not an npm package — skills are plain markdown/code read directly by the agent. Clone the repo, then either run the installer for the whole category or copy the single skill directory.

  • Claude Code — clone and install via the bundled script: 
    git clone https://github.com/GPTomics/bioSkills
cd bioSkills
./install-claude.sh --categories "clinical-databases"

    The installer copies matching skills into ~/.claude/skills/ (default target). Use ./install-claude.sh --list to preview the skills first.

  • Claude Code / other agents — copy just this one skill: 
    cp -r bioSkills/clinical-databases/acmg-classification ~/.claude/skills/

    (run from the directory holding your clone — if you are still in bioSkills/ from the previous step, use cp -r clinical-databases/acmg-classification ~/.claude/skills/, or replace bioSkills/ with the absolute path of your clone). The workflow uses requests for API lookups:

    pip install requests

What it does

Walks the evidence codes and combines them on the Tavtigian 2018/2020 Bayesian point scale, which is the arithmetic behind current ClinGen practice:

  • Loss-of-function — the Abou Tayoun 2018 PVS1 decision tree (exon skipping, NMD escape, last-exon truncation), with AutoPVS1-style automation.
  • In-silico evidence, calibrated — Pejaver 2022 thresholds for REVEL/BayesDel/VEST4 and Bergquist 2025 calibration for AlphaMissense, so PP3/BP4 is applied at a defensible strength rather than as a blanket “predicted damaging”.
  • Splicing — Walker 2023 SpliceAI DS_max bands (≥ 0.2 → PP3_Supporting), with SpliceVault for aberrant-transcript expectations.
  • Functional data — Brnich 2020 OddsPath for setting PS3/BS3 strength from assay validity.
  • Population frequency — ClinGen SVI BA1 (> 5% by default) and the Whiffin 2017 maximum-credible-allele-frequency computation for BS1.
  • Gene-specific rules — VCEP CSpec specifications, which override the generic defaults where they exist.
  • Somatic tiers — AMP/ASCO/CAP 2017 Tier I–IV assignment (Li 2017) for tumour variants, kept separate from germline P/LP/VUS/LB/B.

Primary use cases: germline variant curation, resolving VUS with calibrated evidence, applying VCEP rules for a specific gene, assigning clinical actionability tiers to somatic findings.

Notes

Classification output is research and curation support, not a clinical report. ACMG/AMP classification in a diagnostic setting requires a qualified laboratory director and phenotype context the skill does not have; case-level data (segregation, de novo status, phenotype specificity) has to be supplied by the analyst.

Two practical traps the skill is explicit about: point-scale combination is not interchangeable with the 2015 combining rules in edge cases, and PP3/BP4 must not be applied at full strength from raw predictor output — use the calibrated bands.

Complements the data-source pages it draws on: ClinVar, gnomAD and dbSNP. For somatic interpretation reasoning through ToolUniverse, see Cancer Variant Interpretation.

Distributed as a SKILL.md (plus reference material) in the bioSkills collection. Upstream skill front-matter name is bio-clinical-databases-acmg-classification; upstream directory clinical-databases/acmg-classification. The skill is description-activated — there is no bare /acmg-classification slash command.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.