Docs · OGN platform
GPU-native genomics operating system
From raw reads to GIAB-validated variant calls in a continuous GPU pipeline. This is the control surface for the engine: CLI, pipelines, benchmarks, and deployment runbooks.
CUDA 12+Hopper · AmpereGIAB-validated flowsSchemas stable
Viewing
OGN Core Kit (open source)
OGN Core Kit (open adoption surface)
OGN Core Kit is the public integration surface for Omnis Genome Nexus (OGN): the stable contracts, the worker runner,
SDKs, and workflow adapters. It does not include the proprietary OGN engine binaries or GPU kernels.
What you get
- Job Spec JSON v1 (docs + JSON schema + conformance fixtures)
ogn-runner: stable worker entrypoint (ogn-runner -) that executes an engine and uploads artifacts- Python package
ogn-sdk:ognCLI +ogn_sdkhelpers +ogn_runner - Workflow adapters (example Nextflow pipeline)
Install (Python)
BashPowerShellPython API (coming)
ogn CLIpython -m pip install ogn-sdk
ogn-runner --helpEngine requirement
To run real genomics compute you need access to an OGN engine binary or container image (distributed separately).
ogn-runner shells out to an engine executable (defaults to ogn_run on PATH):BashPowerShellPython API (coming)
ogn CLIexport OGN_ENGINE_BIN=/path/to/ogn_runUsing ogn-runner
Workers are expected to run:
BashPowerShellPython API (coming)
ogn CLIogn-runner -where the Job Spec JSON v1 payload is provided on stdin.
For local testing you can also pass a file path:
BashPowerShellPython API (coming)
ogn CLIogn-runner job_spec.jsonUpload behavior:
- If
outputs.*.put_urlis present, the runner uploads via presigned HTTP PUT. - If
put_urlis absent, the runner uploads directly to the stableoutputs.*.uriwhen supported (e.g.s3://...).
Contracts
The canonical contract lives in the core kit repo:
spec/v1/jobspec.mdspec/v1/jobspec.schema.jsonspec/v1/conformance/
Job Spec v1 is additive-only and consumers must ignore unknown fields.