This repository defines the OmniBioAI Control Center — the orchestration, deployment, and operational control plane of the OmniBioAI ecosystem.
OmniBioAI is a reproducible scientific execution and reasoning platform designed to run consistently across:
- Local machines
- On-prem servers
- HPC environments
- Cloud infrastructure
With no mandatory cloud dependencies.
This repository does not embed bioinformatics algorithms or core application logic. Instead, it assembles independently versioned OmniBioAI components into a single runnable, production-grade ecosystem.
Think of this repository as the operational brain and runtime assembly layer of OmniBioAI.
The Control Center is responsible for:
- Service orchestration
- Environment configuration
- Runtime boundaries
- Cross-service wiring
- Health checks and dependency contracts
- Persistence management
- Deployment portability (local ↔ HPC ↔ cloud)
It defines how OmniBioAI runs — not what each component internally implements.
It is the control plane of the ecosystem.
OmniBioAI follows a multi-plane architecture:
- Django Workbench
- Model Registry
- Tool Execution Service (TES)
- ToolServer
- LIMS
- Metadata databases
- Object and artifact governance
- API boundaries
- Health contracts
These services are:
- Long-lived
- Stateful
- Restart-safe
- Governance-aware
- Workflow runners (WDL, Nextflow, Snakemake, CWL)
- Tool runtime containers
- HPC adapters
- Slurm execution
- Kubernetes jobs
- Cloud batch adapters
These services are:
- Ephemeral
- Replaceable
- Execution-only
TES is the strict boundary between control and compute planes.
- OmniObjects
- Model artifacts
- Workflow outputs
- Versioned bundles
- Datasets
- RAG services
- LLM reasoning
- Agent orchestration
- Scientific interpretation layers
This repository is intended for:
- Core OmniBioAI developers
- Infrastructure engineers
- HPC administrators
- Enterprise deployment teams
- Regulated or air-gapped environments
- Power users running full stacks locally
It is not required for:
- Plugin-only development
- SDK-only usage
- Users consuming hosted OmniBioAI
Desktop/machine/
├── omnibioai/ # Workbench (Django platform)
├── omnibioai-tes/ # Tool Execution Service
├── omnibioai-toolserver/ # FastAPI ToolServer
├── omnibioai-model-registry/ # Model registry service
├── omnibioai-lims/ # LIMS
├── omnibioai-rag/ # RAG & LLM services
├── omnibioai_sdk/ # Python SDK
├── omnibioai-workflow-bundles/ # Engine-agnostic workflows
│
├── deploy/
│ ├── compose/ # Canonical Docker Compose
│ ├── scripts/ # Bootstrap utilities
│ ├── bundle/ # Offline bundles
│ ├── hpc/ # Apptainer assets
│ └── k8s/ # Kubernetes (in progress)
│
├── data/
├── work/
├── tmpdata/
├── out/
├── local_registry/
│
├── db-init/
├── utils/
├── images/
│
├── docker-compose.yml
├── .env.example
└── README.md
| Service | Port | Role |
|---|---|---|
| OmniBioAI Workbench | 8000 | UI, plugins, agents |
| Tool Execution Service (TES) | 8081 | Workflow orchestration |
| ToolServer | 9090 | Tool APIs |
| Model Registry | 8095 | Versioned ML artifacts |
| LIMS | 7000 | Sample & metadata management |
| MySQL | 3306 | Databases |
| Redis | 6379 | Celery & caching |
All ports are configurable via .env.
- Ordered startup with health checks
- Restart-safe dependency management
- Strict port contracts
- Environment-based configuration
- Model versioning
- Alias promotion
- Artifact verification
- Reproducibility metadata
- Strict package validation
Supports execution on:
- Slurm clusters
- Apptainer / Singularity environments
- Non-root HPC nodes
Control plane can remain external while compute runs on HPC.
OCI-compliant images allow deployment to:
- AWS Batch
- Azure Batch
- Kubernetes
- On-prem Docker
Design goal: parity across environments
This repository supports:
- Prebuilt Docker image bundles
- Volume snapshots
- Seeded databases
- Fully offline installation
Suitable for:
- Regulated research labs
- Secure enterprise networks
- Hospital environments
- Docker Engine or Docker Desktop
- Docker Compose v2+
cp deploy/compose/.env.example deploy/compose/.env
docker compose \
--project-directory . \
--env-file deploy/compose/.env \
-f deploy/compose/docker-compose.yml \
up -dcurl http://127.0.0.1:8000
curl http://127.0.0.1:8081/health
curl http://127.0.0.1:8095/health| Mode | Control Plane | Compute Plane |
|---|---|---|
| Local dev | Docker Compose | Local Docker |
| On-prem | Docker Compose | Docker / TES |
| HPC | External VM | Apptainer via TES |
| Hybrid | VM | HPC + TES |
| Cloud | Kubernetes | Kubernetes |
- Single workspace root
- No absolute paths
- Strict service boundaries
- Control plane ≠ compute plane
- Restart-safe orchestration
- Container-native design
- Environment-driven configuration
- No forced cloud dependencies
- Does not contain bioinformatics algorithms
- Does not vendor component repositories
- Does not enforce a single workflow engine
- Does not hide execution behind opaque AI calls
- Does not require external SaaS services
OmniBioAI components are independently versioned repositories.
This repository:
- Assembles them
- Wires them
- Runs them
- Governs them
- Deploys them
It is the operational control boundary of OmniBioAI.
- Multi-service orchestration stable
- Model Registry integrated
- TES integrated
- ToolServer integrated
- LIMS integrated
- Offline-ready structure
- HPC-aware architecture
- Kubernetes preparation underway
This repository is:
The OmniBioAI Control Center — the orchestration, deployment, and runtime control plane of the OmniBioAI ecosystem.
It defines how the ecosystem runs, scales, and moves across environments — while preserving strict reproducibility and architectural boundaries.