Skip to content

Analysis

AIVA provides analysis tools that go beyond data exploration, supporting structured variant interpretation workflows used in clinical and research settings.

In This Section

Tertiary Analysis

Analyze samples with the data table and AI chat side by side. Filter, flag, and interpret variants in a unified workspace.

Tertiary Analysis

ACMG Classification

Apply ACMG/AMP criteria to variants with an interactive evidence-based classifier. Auto-calculated five-tier classifications from Pathogenic to Benign.

ACMG Classification

Analysis Workflow

A typical analysis workflow in AIVA proceeds through the following stages:

  1. Upload and annotate: Upload a VCF file with optional Small Variant Annotation or Structural Variant Annotation.
  2. Explore in the data table: Use the Data Table to browse, filter, and sort variants.
  3. Tertiary analysis: Open the Tertiary Analysis view to work with the data table and AI chat simultaneously.
  4. Flag and comment: Flag variants of interest and add comments documenting your interpretation.
  5. Classify: Apply ACMG criteria to variants requiring formal evidence-based classification.
  6. Export and report: Export your findings and generate clinical reports.

AI-Assisted Analysis

The AIVA assistant supports every stage of analysis:

  • Data queries: Ask AIVA to find specific variants, count categories, or summarize patterns in your data.
  • Literature search: Use AIVA to search biomedical literature, ClinVar, and the web for evidence relevant to specific variants or genes.
  • Knowledge graph: Query the gene-protein-drug interaction graph for pathway analysis and drug-target relationships.
  • Statistical analysis: Ask AIVA to run statistical tests or generate plots using the Code Interpreter.
  • Clinical trials: Search for relevant clinical trials based on genes, conditions, or interventions.

See AI Tools Reference for a complete description of all tools available to AIVA.

Combine tools for comprehensive interpretation

For variant interpretation, a powerful pattern is to ask AIVA a compound question: "Look up ClinVar and gnomAD data for this variant, search for relevant literature, and check for related clinical trials." AIVA chains multiple tools to deliver a comprehensive answer.