Skip to main content
Purna supports ACMG/AMP variant classification with 28 standard criteria. The ACMG tab in the variant detail panel displays active criteria, evidence summaries, and tools for both AI-enhanced and manual classification.

​
ACMG criteria overview

The ACMG/AMP framework uses 28 criteria to classify variants across five categories:
ClassificationCriteria prefixDescription
PathogenicPVS, PSVery strong and strong evidence of pathogenicity
Likely PathogenicPMModerate evidence of pathogenicity
SupportingPP, BPSupporting evidence for or against pathogenicity
Likely BenignBSStrong evidence of benign impact
BenignBAStand-alone evidence of benign impact

​
Viewing ACMG criteria

Open the ACMG tab in the variant detail panel to see:
  • Classification badge — The current overall classification (Pathogenic, Likely Pathogenic, VUS, Likely Benign, or Benign)
  • Active criteria — Each criterion that applies to this variant, shown as a labeled badge (e.g., “BA1”, “BS1”, “BP7”)
  • Evidence summaries — Brief descriptions of why each criterion was applied
  • Computational evidence — Aggregated results from SIFT, PolyPhen-2, AlphaMissense, and CADD
Each criterion badge indicates whether it was determined by computational analysis (DET) or by AI review (AI). ACMG tab

​
AI-enhanced classification

Click the AI Enhance button to trigger an AI-powered evaluation of ACMG criteria. This is a three-step workflow:

​
Step 1: Start

Click AI Enhance to begin. Purna initiates the analysis and begins gathering evidence.

​
Step 2: Running

Purna evaluates the variant against ACMG criteria using evidence from multiple sources:
  • ClinVar — Clinical significance submissions and review status
  • PubMed — Published literature on the variant and gene
  • gnomAD — Population allele frequency data
  • Functional data — In silico predictions and conservation scores
  • Phen2Gene — Phenotype-gene association ranking
A progress indicator shows which evidence sources are being queried.

​
Step 3: Results

The AI presents its findings:
  • Suggested criteria — ACMG criteria the AI recommends adding or modifying
  • Confidence scores — How confident the AI is in each suggestion
  • Evidence citations — Sources supporting each criterion
  • DET vs AI badges — Distinguishing computational determinations from AI-suggested criteria
Review each suggestion and choose to accept or reject individual criteria. Accepted criteria update the variant’s ACMG classification.
AI-enhanced classification is a decision-support tool. Always review the evidence and apply your clinical judgment before accepting AI suggestions.

​
Manual reclassification

Click the Reclassify button to manually update a variant’s classification:
  1. Select the new classification from the dropdown (Pathogenic, Likely Pathogenic, VUS, Likely Benign, or Benign)
  2. Add criteria that support your reclassification
  3. Provide a justification note explaining your reasoning
  4. Click Save to apply the reclassification

​
Reclassification history

Every reclassification is recorded in the variant’s history. The history shows:
  • Date of each reclassification
  • Previous classification and new classification
  • Who made the change
  • Justification provided
You can revert to a previous classification if needed.
Document your reasoning when reclassifying variants. The reclassification history creates an audit trail that supports reproducible interpretation and peer review.