Skip to content

Variant Classification

The variant classifier is a default category card on the Analysis Panel that appears for all samples. When reviewing a variant in the analysis panel, click Classify Variant to open the classifier. AIVA supports both ACMG/AMP germline and AMP/ASCO/CAP somatic classification frameworks.

Classification Frameworks

ACMG/AMP (Germline)

The standard five-tier classification for germline variants:

Classification Meaning
Pathogenic The variant is disease-causing with strong supporting evidence.
Likely Pathogenic The variant is probably disease-causing (>90% certainty).
VUS Insufficient evidence to classify definitively.
Likely Benign The variant is probably not disease-causing (>90% certainty).
Benign The variant is non-pathogenic with strong supporting evidence.

AMP/ASCO/CAP (Somatic)

The four-tier classification for somatic variants:

Tier Meaning
Tier I Strong clinical significance: FDA-approved therapies or professional guidelines.
Tier II Potential clinical significance: investigational therapies or clinical trials.
Tier III Unknown clinical significance: no established evidence.
Tier IV Benign or likely benign: no known oncogenic role.

Classification Options

When you click Classify Variant, you have two options:

Manual classification

Select criteria yourself based on your review of the evidence:

  1. The classifier displays all applicable criteria organized by category.
  2. Check the criteria that are met. You can select any combination of pathogenic and benign criteria.
  3. For each selected criterion, add a note documenting the specific evidence (e.g., "gnomAD AF = 0.0001, absent in controls").
  4. The classification auto-calculates in real time as you add or remove criteria.
  5. Review and click Save.

AI classification

Let AIVA classify the variant automatically:

  1. Click AI Classify to send the variant to an AI sub-agent.
  2. The sub-agent analyzes both the patient's phenotype and the variant's annotations (allele frequency, in silico predictions, ClinVar data, literature) to determine applicable criteria.
  3. Results are returned with the suggested classification, selected criteria, and sources for each piece of evidence used.
  4. Review the AI's reasoning and sources, adjust any criteria if needed, then click Save.

AI classification is a starting point

Always review the AI's suggested criteria and sources before saving. The AI provides a thorough initial assessment, but clinical judgment should guide the final classification.

Saving and Visibility

  1. Review the selected criteria and the auto-calculated classification.
  2. Click Save to store the classification.
  3. The classification is associated with the variant and visible in the table view.

Tips

  • Document your reasoning: Add notes to each selected criterion explaining the specific evidence.
  • Revisit VUS variants: Variants classified as VUS should be periodically re-evaluated as new evidence becomes available.