GPU-native genomics

Request Snapshot Report

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

Pilot quickstart (Compose stack)

Benchmark pack Example pipelines GitHub

Pilot quickstart (Docker Compose demo)

This quickstart is a self-contained pilot demo you can run on your laptop. It simulates the OGN “submit → run → download artifacts” flow and returns a small proof bundle (metrics JSON + a tiny VCF).

It is intentionally lightweight: no GPUs, no real genomics compute. Snapshot Reports replace this demo server with the GPU pipeline and return your real artifacts.

1. Requirements

Docker Desktop / Docker Engine
Docker Compose v2 (docker compose)
curl (and optionally jq)

2. Download the compose stack

Create a clean working directory:

BashPowerShellPython API (coming)

ogn CLI

mkdir -p ogn-pilot-demo && cd ogn-pilot-demo

Download the compose file and demo server:

BashPowerShellPython API (coming)

ogn CLI

curl -L -o docker-compose.yml https://omnisgenomics.com/pilot-demo/docker-compose.yml
curl -L -o demo_server.py https://omnisgenomics.com/pilot-demo/demo_server.py

3. Start the stack

BashPowerShellPython API (coming)

ogn CLI

docker compose up

You should see the server listening on http://localhost:8080.

Health check:

BashPowerShellPython API (coming)

ogn CLI

curl -s http://localhost:8080/healthz

4. Submit a run

Submit a demo run (any JSON payload is accepted):

BashPowerShellPython API (coming)

ogn CLI

curl -sS -X POST http://localhost:8080/v1/runs \
  -H 'Content-Type: application/json' \
  -d '{
    "pipeline": "wgs30",
    "sample": "hg002",
    "reads": ["R1.fastq.gz", "R2.fastq.gz"],
    "ref": "GRCh38_noalt.fa"
  }'

If you have jq, extract the run id:

BashPowerShellPython API (coming)

ogn CLI

RUN_ID=$(curl -sS -X POST http://localhost:8080/v1/runs -H 'Content-Type: application/json' -d '{"sample":"hg002"}' | jq -r .id)
echo "$RUN_ID"

5. Poll status and download artifacts

Poll until the run reaches succeeded (the demo returns quickly with synthetic artifacts):

BashPowerShellPython API (coming)

ogn CLI

curl -sS http://localhost:8080/v1/runs/$RUN_ID

List available artifacts:

BashPowerShellPython API (coming)

ogn CLI

curl -sS http://localhost:8080/v1/runs/$RUN_ID/artifacts

Download the proof bundle artifacts:

BashPowerShellPython API (coming)

ogn CLI

curl -sS -o metrics.json http://localhost:8080/v1/runs/$RUN_ID/artifacts/metrics.json
curl -sS -o variants.vcf http://localhost:8080/v1/runs/$RUN_ID/artifacts/variants.vcf
curl -sS -o proof_bundle.json http://localhost:8080/v1/runs/$RUN_ID/artifacts/proof_bundle.json

6. Tear down

BashPowerShellPython API (coming)

ogn CLI

docker compose down

Pilot quickstart (Compose stack) | OGN documentation | Omnis Genomics