tooluniverse-acmg-variant-classification

Classifies germline variants using ACMG/AMP criteria with evidence from ClinVar, gnomAD, and computational predictors.

1.7k|254|Updated Mar 3, 2025

Search Tags:#clinvar #variant classification #acmg #gnomad #pathogenicity #clinical genetics #vus

Ready to run

npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-acmg-variant-classification

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: tooluniverse-acmg-variant-classification
Source: https://github.com/mims-harvard/ToolUniverse/tree/main/plugins/tooluniverse/skills/tooluniverse-acmg-variant-classification
Command: npx skills add https://github.com/mims-harvard/ToolUniverse --skill tooluniverse-acmg-variant-classification

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Interpreting the clinical significance of genetic variants requires systematically applying 28 ACMG/AMP criteria across population frequencies, computational predictions, clinical databases, and literature, which is error-prone and time-consuming when done manually.

Core Features & Use Cases

Systematic 28-Criteria Evaluation: Applies all ACMG/AMP criteria (PVS1, PS1-4, PM1-6, PP1-5, BA1, BS1-4, BP1-7) with cited evidence per criterion and conservative strength assignment.
Multi-Source Evidence Gathering: Queries ClinVar, gnomAD, MyVariant, Ensembl VEP, UniProt, InterPro, AlphaFold, and PubMed through ToolUniverse tools to ground every criterion activation.
Five-Tier Classification Output: Produces a Pathogenic / Likely Pathogenic / VUS / Likely Benign / Benign verdict with a structured report showing evidence tables and classification logic.
Use Case: A clinical geneticist asks "Classify BRCA2 c.5946delT using ACMG criteria" and receives a full report combining PVS1 (frameshift in LOF gene), ClinVar entries, gnomAD absence, and literature evidence into a Pathogenic verdict.

Quick Start

Ask the agent to classify a variant such as "Apply ACMG guidelines to NM_000059.4:c.5946delT and give me the pathogenicity verdict with supporting evidence."

Frequently Asked Questions about tooluniverse-acmg-variant-classification

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I classify a genetic variant using ACMG criteria?▼

Provide the variant as HGVS notation, rsID, or gene plus change, and the workflow validates it against the MANE Select transcript, then gathers population, computational, clinical, and literature evidence. The output is a five-tier verdict with each activated criterion cited to its source database.

What databases are used for variant pathogenicity assessment?▼

The workflow queries ClinVar for clinical submissions, gnomAD for ancestry-specific population frequencies, MyVariant and Ensembl VEP for predictors like REVEL, CADD, and SpliceAI, plus UniProt, InterPro, AlphaFold, and PubMed for functional context.

Can this resolve a variant of uncertain significance (VUS)?▼

Yes, VUS resolution is a primary use case. The workflow combines criteria at their applied strengths using the ACMG combination rules, and can add ESM-based mechanistic explanations showing which protein features a missense variant disrupts.

Does ACMG classification work for non-missense variants?▼

Yes, frameshift, nonsense, splice-site, synonymous, and in-frame indel variants are supported. Missense-specific criteria like PP3 and BP4 are skipped for other types, and splice impact is assessed through SpliceAI scores and PVS1 rules.

What are the limitations of automated ACMG classification?▼

Criteria requiring patient clinical data such as de novo occurrence (PS2), segregation (PP1 beyond literature), and phenotype specificity (PP4) cannot be assessed automatically and are marked Not Assessed. The classification also depends on database coverage, so absent gnomAD data is treated as a gap, not evidence.