Kaviar (~Known VARiants) is a compilation of SNVs, indels, and complex variants observed in humans, designed to facilitate testing for the novelty and frequency of observed variants.


  • Released: February 29, 2016 (version 160204-Public)
  • Kaviar contains 162 million SNV sites (including 25M not in dbSNP) and incorporates data from 35 projects encompassing 77,781 individuals (13.2K whole genome, 64.6K exome).
  • Kaviar also contains 50 million short indels and substitutions from a subset of the data sources.
  • Kaviar excludes cancer genomes but includes some data from cell lines and individuals affected by disease.
  • An effort was made to exclude data from related individuals.
  • Release notes, known issues, and changelog
  • Auxiliary info for some data sources.
  • Treemap depiction of sources.
  • Caution: variant novelty need not imply functional or clinical relevance. Conversely, known variants may have unknown functional implications. In some cases, even the reference form may be the detrimental variant.

Using it online

  • Query Kaviar interactively by clicking the Query Kaviar button above. Appropriate for small to medium size queries.
  • Query Kaviar as a web service using this URL in your code (here is example usage).
  • Query Kaviar via its beacon, as encouraged by the GA4GH.
Note: options in this box only appropriate if your data are not restricted (e.g., by IRB protocols) to prevent use of web services for analysis.