Skip to content

Pillar I: Data

GKM describes one piece of genomic knowledge, such as a variant, a category of variants, or a clinical assertion, in a consistent way. Real resources contain many of these objects and need consistent ways to share them. This pillar shows how GKM data can be packaged and shared in practice, from individual records to bundles, GKM's compact record format, and larger datasets.

Ways to share GKM data

The initial implementation supports some sharing patterns today. Others remain in development or are future directions. The right delivery method depends on the scale and use case.

  • Native

    Available now

    A single record or a small set. Each object is self-contained, with nothing compacted.

    Formats: JSON, JSONL, YAML

  • Bundle

    Available now

    GKM's format for compact records. Packages related objects together, stating shared representations once and linking to them by reference.

    Formats: JSON, JSON Schema

  • Bulk

    In development

    Bulk formats such as Parquet or relational tables for very large collections, as support develops.

Native and bundled representations

Native records carry related objects inline. Bundles place shared objects in collections and link to them with bundle-local JSON Pointers. Below, two alleles share a location but have different states. Native records repeat the location and sequence reference; the bundle stores them once.

[
  {
    "type": "Allele",
    "location": {
      "type": "SequenceLocation",
      "sequenceReference": {
        "type": "SequenceReference",
        "refgetAccession": "SQ.dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7"
      },
      "start": 43093453,
      "end": 43093454
    },
    "state": {
      "type": "LiteralSequenceExpression",
      "sequence": "C"
    }
  },
  {
    "type": "Allele",
    "location": {
      "type": "SequenceLocation",
      "sequenceReference": {
        "type": "SequenceReference",
        "refgetAccession": "SQ.dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7"
      },
      "start": 43093453,
      "end": 43093454
    },
    "state": {
      "type": "LiteralSequenceExpression",
      "sequence": "T"
    }
  }
]

{
  "sequenceReference": {
    "SQ.dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7": {
      "type": "SequenceReference",
      "refgetAccession": "SQ.dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7"
    }
  },
  "location": {
    "location:1": {
      "type": "SequenceLocation",
      "sequenceReference": "#/sequenceReference/SQ.dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7",
      "start": 43093453,
      "end": 43093454
    }
  },
  "allele": {
    "allele:1": {
      "type": "Allele",
      "location": "#/location/location:1",
      "state": {
        "type": "LiteralSequenceExpression",
        "sequence": "C"
      }
    },
    "allele:2": {
      "type": "Allele",
      "location": "#/location/location:1",
      "state": {
        "type": "LiteralSequenceExpression",
        "sequence": "T"
      }
    }
  }
}

Built with community partners

The Starter Kit uses real content developed with resources such as ClinVar GKM and CIViC. Each partner defines its collection names and object groupings in a bundle schema, preserving familiar terminology while using shared GKM models.

This section will grow with the community's needs.