VarScore Variant Scoring System
AI-driven clinical variant interpretation: auto-reads literature, extracts evidence, ACMG/AMP classification, full SNP·SV·ROH coverage.
What VarScore Variant Scoring System does
VarScore is an AI-driven, commercial clinical variant interpretation platform benchmarked against Franklin / Genoox core workflows. Its most cutting-edge capability lets AI automatically search and read vast medical literature, extract structured pathogenicity evidence and map it to ACMG criteria — turning the most time-consuming literature review into minute-level assistance. It provides ACMG/AMP classification across SNP, structural variant (SV) and runs-of-homozygosity (ROH) workflows, with local Ensembl VEP annotation, batch VCF analysis and PDF reports.
Built for real-world scenarios
AI literature reading & evidence extraction
The core, cutting-edge capability: AI auto-searches and reads relevant literature, extracts functional, segregation, case and de novo evidence, and maps it to ACMG criteria (PS3/BS3, PP1, PS2) — turning hours of review into minutes.
Interactive ACMG/AMP classification
Following Richards 2015: all 28 SNP ACMG criteria and ClinGen 2020 CNV sections 1–5, with per-criterion toggles, strength adjustment and real-time reclassification.
SNP · SV · ROH workflows
Three dedicated templates: SNP (ACMG-first), SV (ACMG + region viewer), ROH (genes & regions first).
6-track region viewer
Genes, ClinGen dosage sensitivity, ClinVar, DGV Gold, DGV and gnomAD-SV multi-track visualization in genomic context.
Multi-predictor evidence
Aggregates REVEL, AlphaMissense, SpliceAI, EVE, PrimateAI plus gnomAD frequencies and gene constraint (pLI/LOEUF).
Batch analysis & reports
Up to 200-variant batch VCF analysis, classification persistence, community consensus and PDF report export.
One variant, five steps to sourced evidence
Hand the clinician's most time-consuming task — read papers, find evidence, map criteria — to AI; every piece of evidence traces back to its source.
AI summary: ATPase activity and actin binding significantly reduced. Literature [example]
AI summary: parents non-carriers, relationships verified. Case report [example]
AI summary: 3 affected family members all carry it. Family study [example]
AI check: absent in gnomAD, meets PM2. gnomAD (auto)
[Example data, capability demo only] Every piece of evidence links to its source, supporting manual review, criterion-strength adjustment and real-time reclassification — AI handles the reading and searching; the call stays with the expert.
See it in action
Key advantages
About VarScore Variant Scoring System
What is the VarScore Variant Scoring System?
VarScore is an AI-driven, commercial clinical variant interpretation platform benchmarked against Franklin/Genoox, providing ACMG/AMP classification across SNP, structural variant (SV) and runs-of-homozygosity (ROH), with AI literature reading and evidence extraction as its core capability.
What exactly does the AI literature-evidence feature do?
For a given variant, VarScore's AI automatically searches and reads relevant medical literature, extracts functional studies, co-segregation, case reports and de novo evidence, and maps it to the corresponding ACMG criteria (e.g., PS3/BS3, PP1, PS2), with traceable citations — shrinking hours of review to minutes.
Which variant types and ACMG criteria are supported?
SNP interpretation implements all 28 ACMG 2015 criteria (incl. PVS1, PM1, PS2/PM6, PP1/BS4, BP7); structural variants follow ClinGen 2020 CNV sections 1–5; ROH uses a genes-and-regions-first layout.
Which predictors and reference data are integrated?
REVEL, AlphaMissense, SpliceAI, EVE, PrimateAI, dbscSNV and more, plus ClinVar, gnomAD, DGV, dbNSFP, GTEx and ClinGen reference data and population frequencies.
Which query formats and batch analysis are supported?
Genomic coordinates (chr9-135800978-A-G), HGVS, rsID, SV coordinates and array notation, plus up to 200-variant batch VCF analysis and PDF report export.
Learn more about VarScore Variant Scoring System
Learn more about VarScore, or browse the full MeiTian product matrix.