Open-source AI platform for genomics, bioinformatics, biomedical research,
and evidence-based AI powered by public scientific databases and multi-provider LLMs.
Try GenomeAI right now — no install required. Click the link and use the working application:
🌐 Open the Live Demo → https://genomeai.vercel.app
Backend API: https://genomeai-api.onrender.com · free-tier hosts, deployed automatically on every release.
See docs/deployment/releases.md for how it works.
GenomeAI is a unified platform that brings together public biological databases, workflow automation, data visualization, and AI-powered analysis — all in one open-source stack.
It connects directly to 18 free public APIs (NCBI, Ensembl, UniProt, ClinVar, gnomAD, PDB, AlphaFold, ChEMBL, PubChem, Reactome, KEGG, STRING, OpenTargets, Monarch, Disease Ontology, DGIdb, Europe PMC, Semantic Scholar) to give researchers instant access to genomic, proteomic, chemical, and clinical data — without expensive licenses or proprietary infrastructure.
Goal: Make biomedical research accessible, reproducible, and AI-augmented for every researcher, clinician, and developer — anywhere in the world.
| Problem | GenomeAI Solution |
|---|---|
| Biomedical data is scattered across 100+ databases | Unified connector layer — one API for all sources |
| Existing tools require expensive licenses | 100% open-source, runs on free-tier APIs |
| No integrated AI for genomic analysis | Multi-provider LLM support (Gemini cloud, Ollama local) |
| Workflows are manual and error-prone | DAG-based workflow engine with retry, scheduling, and parallel execution |
| Results are hard to visualize | Built-in genome browser, protein viewer, network graphs, scientific charts |
| Reproducibility is difficult | Versioned workflows, provenance tracking, audit logs |
| Feature | Status |
|---|---|
| Biological domain models (Genome, Sample, Gene, Variant, Transcript, Protein, …) | ✅ |
| REST API with CRUD + enhanced analysis endpoints (24 route modules) | ✅ |
| Workflow DAG engine (definitions, steps, dependencies, validation) | ✅ |
| Deterministic, parallel DAG execution | ✅ |
| Cron-based workflow scheduler | ✅ |
| Queue & worker (Redis-backed background execution) | ✅ |
| Retry & failure handling (classification, backoff, policies) | ✅ |
| 18 real external-data connectors (NCBI, UniProt, ClinVar, gnomAD, …) | ✅ |
| AI-powered analysis (gene, variant, protein, drug, pathway, disease, literature, multi-domain report) with Gemini + Ollama | ✅ |
| Genome browser, protein viewer, network graphs, scientific charts, molecular structure | ✅ |
| Live demo — genomeai.vercel.app | ✅ |
| Feature | Phase |
|---|---|
| Knowledge graph (gene-disease-drug-protein associations) | 5 |
| More AI providers (OpenAI, Anthropic, Groq, Mistral) | 6 |
| PubMed/Europe PMC deep-dive literature QA | 6 |
| Plugin SDK and marketplace | 7 |
| Full-text ingestion workers (scheduled sync of source databases) | 7 |
| Authentication & multi-user organizations | 8 |
GenomeAI ships with 18 real connectors to free public databases (implemented, live, and tested):
| Domain | Sources |
|---|---|
| Genomic | NCBI E-utilities, Ensembl (VEP) |
| Protein | UniProt, RCSB PDB, AlphaFold DB |
| Drug/Compound | ChEMBL, PubChem, DGIdb, Open Targets |
| Literature | Europe PMC, Semantic Scholar |
| Pathway | Reactome, KEGG |
| Network | STRING |
| Variant | ClinVar, gnomAD |
| Disease | Disease Ontology, Monarch, Open Targets |
See docs/data-integration/ for the live connectors and docs/external-data/MASTER_PLAN.md for the historical full integration plan (including future sources).
┌──────────────────────────────────────────────────────────────────┐
│ User Interfaces │
│ CLI | REST API | Web UI | Python SDK │
└──────────────────────────────┬───────────────────────────────────┘
│
┌──────────────────────────────┴───────────────────────────────────┐
│ API Gateway (FastAPI) │
│ Auth | Rate Limiting | Validation | Audit │
└──────────────────────────────┬───────────────────────────────────┘
│
┌──────────────────────────────┴───────────────────────────────────┐
│ Orchestration Layer │
│ Workflow DAG Engine | Cron Scheduler | Redis Queue Worker │
│ (parallel exec) | (scheduling) | (retry, backoff) │
└──────────────────────────────┬───────────────────────────────────┘
│
┌──────────────┬───────────────┼───────────────┬───────────────────┐
│ │ │ │ │
│ Biological │ Data │ AI/LLM │ Visualization │
│ Domains │ Integration │ Gateway │ Platform │
│ │ │ │ │
│ Genome │ NCBI │ Gemini │ Genome Browser │
│ Sample │ Ensembl VEP │ Ollama │ Protein Viewer │
│ Gene │ UniProt │ (local) │ Network Graphs │
│ Variant │ ClinVar │ │ Scientific │
│ Transcript │ gnomAD │ │ Charts │
│ Protein │ PDB/AlphaFold │ │ 3D Molecular │
│ │ ChEMBL/PubChem │ │ │
│ │ Reactome/KEGG │ │ │
│ │ STRING/OpenTargets│ │ │
│ │ EuropePMC/S2 │ │ │
└──────────────┴───────────────┴───────────────┴───────────────────┘
│
┌──────────────────────────────┴───────────────────────────────────┐
│ Storage Layer │
│ PostgreSQL | Redis | Object Storage │
└──────────────────────────────────────────────────────────────────┘
See ARCHITECTURE.md for the full technical breakdown.
# Clone the repository
git clone https://github.com/dsk-dev-ai/GenomeAI.git
cd GenomeAI
# Install all dependencies and start infrastructure
make setup
# Run tests (over 2,150 passing)
make test
# Start the API server
uvicorn genomeai_api.main:app --reloadRequirements: Python 3.12+, Node.js 20+, PostgreSQL 16+, Redis 7+
See docs/development/ for detailed setup guides.
| Layer | Technology |
|---|---|
| API | Python 3.12, FastAPI, SQLAlchemy, Pydantic |
| Database | PostgreSQL 16+, Redis 7+ |
| Workflow Engine | Python asyncio, DAG execution, Redis queue |
| Frontend | Next.js, React, TypeScript, Tailwind CSS |
| Visualization | D3.js, Three.js, Cytoscape.js, Mol* |
| AI/LLM | Gemini (cloud, default gemini-3.6-flash), Ollama (local) |
| DevOps | Docker, GitHub Actions, Turbo (monorepo) |
| Testing | pytest (2,150+ tests), biome, ruff, pyright |
| Topic | Location |
|---|---|
| Architecture | ARCHITECTURE.md |
| Roadmap | ROADMAP.md |
| API Reference | docs/api/ |
| Database Schema | docs/database/ |
| Workflow Engine | docs/workflows/ |
| External Data & APIs | docs/external-data/ |
| AI/ML Guide | docs/ai/ |
| Visualization | docs/visualization/ |
| Decisions | docs/decisions/ |
| Contributing | CONTRIBUTING.md |
We welcome contributions from bioinformaticians, software engineers, data scientists, and researchers.
See CONTRIBUTING.md for guidelines.
# Quick contribution workflow
git checkout -b feat/my-feature
make lint make typecheck make test
git commit -m "feat(domain): add my feature"
git push origin feat/my-feature
# Open a Pull Request- Live demo: genomeai.vercel.app
- Discussions: GitHub Discussions — Q&A, ideas, show & tell
- Issues: Report a bug · Request a feature
- Code of conduct: CODE_OF_CONDUCT.md
- Security: SECURITY.md
⭐ Live chart — updates automatically as the repo gains stars. Click the graphic for the interactive view (GitHub restricts public star data, so the embed uses an encrypted star-history.com sealed token).
GenomeAI is built and maintained by Darshan Kachare through NextGenAI Labs.
Sponsorship supports development infrastructure, documentation, and long-term maintenance of this open-source platform.
GenomeAI is open source under the Apache 2.0 License.
Built with care for the global research community.
GitHub · Discussions · Sponsor