Skip to content

Using the Variant Classifier

The AIVA Variant Classifier is a public tool for classifying variants according to ACMG/AMP guidelines. No login is required.

The classifier follows the ACMG/AMP framework as described in the 2015 joint consensus recommendation:

Richards S, Aziz N, Bale S, et al. "Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology." Genet Med. 2015;17(5):405-424.

For the full criteria definitions, refer to the original publication or the ClinGen Sequence Variant Interpretation resources.

Gene-specific guidelines

Several genes and gene groups have published modifications to the standard ACMG/AMP criteria (e.g., TP53, RASopathy genes, CDH1, PTEN). When available, gene-specific guidelines should take precedence over the general framework.


Step 1: Access the Classifier

Navigate to the Variant Classifier from the AIVA homepage or public navigation menu. The classifier is accessible without an account.


Step 2: Enter Variant Information

Provide information about the variant you want to classify:

  • Gene: The gene symbol (e.g., BRCA1, TP53, MLH1).
  • Variant notation: HGVS coding or protein notation (e.g., c.5266dupC, p.Gln1756Profs*74).
  • Genomic coordinates (optional): Chromosome, position, reference allele, and alternate allele.
  • Transcript (optional): The reference transcript identifier (e.g., NM_007294.4).

Variant identifiers

You can enter the variant in any commonly used format. The classifier accepts HGVS nomenclature, genomic coordinates (GRCh37 or GRCh38), and rsIDs.


Step 3: Select ACMG/AMP Criteria

The classifier presents the full list of ACMG/AMP evidence criteria organized by type:

Pathogenic Criteria

Category Criteria Description
Very Strong PVS1 Null variant in a gene where loss of function is a known mechanism of disease
Strong PS1--PS4 Same amino acid change, functional studies, segregation, prevalence in affected
Moderate PM1--PM6 Hot spot, absent from controls, protein length change, novel missense, assumed de novo, in-frame in non-repeat
Supporting PP1--PP5 Co-segregation, computational, phenotype specificity, reputable source, missense in low-rate gene

Benign Criteria

Category Criteria Description
Stand-Alone BA1 Allele frequency above 5% in population databases
Strong BS1--BS4 Frequency greater than expected, observed in healthy adults, functional studies, lack of segregation
Supporting BP1--BP7 Missense in truncation gene, observed in trans, in-frame in benign region, computational, synonymous, reputable source

For each criterion:

  1. Review the criterion description to determine whether it applies to your variant.
  2. Click the criterion to select or deselect it.
  3. Selected criteria are highlighted and contribute to the classification calculation.

Step 4: Review the Classification

As you select criteria, the classifier automatically calculates the classification in real time. The current classification is displayed prominently: Pathogenic, Likely Pathogenic, VUS, Likely Benign, or Benign.


Step 5: Document Your Reasoning (Optional)

The classifier allows you to add notes for each selected criterion, documenting the evidence that supports your selection. This is optional for the public tool but recommended for thorough variant interpretation.


Tips for Accurate Classification

  • Apply criteria conservatively: Only select a criterion when the evidence clearly supports it.
  • Use multiple evidence types: A robust classification draws on population data, computational predictions, functional studies, and clinical observations.
  • Consider conflicting evidence: If both pathogenic and benign criteria apply, the variant may be classified as VUS pending additional evidence.

Classification responsibility

The Variant Classifier is a decision-support tool. Clinical variant classifications should be performed by qualified professionals in the context of established laboratory protocols and clinical information.