Bundles

GKM bundles are compact records that package related genomic knowledge for sharing, validation, and reuse across resources. They keep standard GKM objects connected while preserving each producer's organization and context, giving producers flexibility in how they structure their data.

What a bundle looks like

A bundle groups related GKM objects into named collections. A producer defines the collections and their contents in a bundle schema. Relationships between objects in the same bundle use bundle-local JSON Pointers (#/...).

This example has required metadata, a shared data-release extension, and two samples. Both samples point to the same extension, so its value appears once.

{
  "metadata": {
    "createdAt": "2026-09-21",
    "producer": "Example knowledgebase",
    "bundleVersion": "1.0"
  },
  "extensions": {
    "data-release": {
      "type": "Extension",
      "name": "dataRelease",
      "value": "2026-09"
    }
  },
  "samples": {
    "sample:1": {
      "id": "sample:1",
      "type": "Sample",
      "name": "Tumor sample",
      "dataRelease": "#/extensions/data-release"
    },
    "sample:2": {
      "id": "sample:2",
      "type": "Sample",
      "name": "Normal sample",
      "dataRelease": "#/extensions/data-release"
    }
  }
}

Each shared GKM object appears once in a bundle. Other objects link to it with a JSON Pointer instead of repeating it. Bundles are distributed with their bundle schema, which defines the collection names, allowed objects, metadata, and local references.

Continue exploring