Build GA4GH Data Connect API

[ifokkema]

Build GA4GH Data Connect API

LOVD3 needs a new API for full data sharing. Several solutions and data formats were considered, and the GA4GH Data Connect Table API solution was chosen in combination with a VarioML-JSON payload.

Tasks

• Until all details have been worked out, this API requires authentication.
• Handle GET.
• Handle HEAD.
• Add /api/v2/ga4gh endpoint, forwarding to the Table API.
• Add /api/v2/ga4gh/tables as a starting point for the table discovery.
• List the variants table, using a VarioML-JSON-based data model.
• Add /api/v2/ga4gh/table/variants/info URL.
• Add /api/v2/ga4gh/table/variants/data, which should just refer to the first data page.
• Data pages (pagination) should follow the .../data:hg19:chr1:123456 format. This supports using static files.
• Make sure that non-public entries will never be included.
• Make sure that non-public fields will never be included, so check those settings.
• Data without a license or with a non-compatible license should only share summary data.
• Data should be able to be filtered on;
  • License (postponed, optional filter when authorized, take note of Set out rules for cache-friendly conditional requests ga4gh-discovery/data-connect#107)
    • Filter excluding detailed data for BY-ND;0 and BY-NC-ND;0 should be always on, as this data can not be aggregated and shared by us (treat like not having a license at all - share summary data only).
    • LOVD/varcache should identify itself to unhide ND licenses with additional permissions.
    • When LOVD/varcache is not querying, also BY-ND;1 and BY-NC-ND;1 licensed data should not be shared in detail. Summary data only.
  • Date last edited (optional filter using the If-Modified-Since HTTP header, take note of Set out rules for cache-friendly conditional requests ga4gh-discovery/data-connect#107)
• Add /service-info URL.
• Make sure the VarioML payload validates against the JSON Schema file (still to be developed for variant-centered views).
  • Develop the JSON schema for variant-centered views: Create JSON Schema for variant-centered views. VarioML/VarioML#20
• Make sure the whole output validates according to the GA4GH standards.
  • Find a JSON Schema for that, or request one.
• Since this development has taken quite some time, revalidate when all is done and make sure we comply to v.1.0.0.
  • Perhaps we should put the version somewhere?