Local-first AI research partner for multi-omics analysis
Chat with your workflows · run reproducible skills · keep data local · resume with memory
English · 简体中文 · Why · Quick Start · Capabilities · Domains · Docs Site
OmicsClaw turns local multi-omics tools into AI-callable skills. The LLM plans and operates; Python/R/CLI tools process data in your local or remote runtime.
One workspace for chat, datasets, skills, execution, memory, and analysis outputs.
📥 Download the OmicsClaw Desktop App · All releases · SHA256SUMS
The Releases tab hosts the prebuilt desktop installers — same oc desktop-server the CLI ships, wrapped in a chat-ready Electron UI. Pick the asset for your platform:
| Platform | Installer |
|---|---|
 |
OmicsClaw-<ver>-arm64.dmg |
|
OmicsClaw-<ver>-x64.dmg |
 |
OmicsClaw.Setup.<ver>-x64.exe · OmicsClaw.Setup.<ver>-arm64.exe |
 |
.AppImage · .deb · .rpm |
|
.AppImage |
Verify each download against
SHA256SUMS.txtpublished alongside the installers. The desktop client and the CLI talk to the same backend — analyses, memory, and remote runtimes stay portable across both.
| Common pain | OmicsClaw answer |
|---|---|
| Analyses restart from zero | Persistent workspace, sessions, and graph memory |
| Python, R, and CLI tools are scattered | Unified skill runner plus natural-language routing |
| Large data lives on servers | Local UI with remote Linux execution over SSH |
| Reports, artifacts, and parameters drift | Standard skill output contracts and reproducible demos |
| 🧠 Memory Sessions, preferences, lineage |
🔒 Local-first Raw data stays in your runtime |
🧰 89 skills Generated catalog + demos |
🧭 Smart routing Natural language to tools |
🖥️ CLI Surfaceoc interactive, oc tui |
🌐 Desktop Surface FastAPI for desktop/web |
📨 Channel Surface 10 IM adapters (Telegram, Feishu, …) |
📡 Remote mode SSH tunnel to Linux servers |
git clone https://github.com/TianGzlab/OmicsClaw.git
cd OmicsClaw
bash 0_setup_env.sh
conda activate OmicsClaw
oc list
oc run spatial-preprocess --demo --output /tmp/omicsclaw_demoConfigure chat and runtime settings:
oc onboard
oc interactiveIf oc is not on PATH, use python omicsclaw.py <command>.
Three Surfaces, one agent loop. All entries below dispatch into the same backend (see ADR 0005).
| Surface | Entry point | Use it for |
|---|---|---|
| 💬 CLI Surface | oc interactive / oc tui |
Natural-language workflows in the terminal (REPL + full-screen TUI) |
| 🌐 Desktop Surface | oc desktop-server |
FastAPI backend consumed by OmicsClaw-App and browser frontends |
| 📨 Channel Surface | python -m omicsclaw.surfaces.channels --channels <names> |
Telegram, Feishu, Slack, Discord, WeChat, WeCom, DingTalk, iMessage, Email, QQ |
| 🧪 Skill runner (non-Surface) | oc run <skill> --demo |
Reproducible one-shot analysis |
| 🔌 MCP (non-Surface) | oc mcp add ... |
External tool integration |
| 📡 Remote mode | oc desktop-server over SSH |
Server-side data and jobs |
Remote mode uses 127.0.0.1, SSH tunneling, and OMICSCLAW_REMOTE_AUTH_TOKEN. See remote execution and the legacy remote guide.
| Path | Best for | Command |
|---|---|---|
| 🥇 Full conda | Real analysis with Python + R + bioinformatics CLIs | bash 0_setup_env.sh |
| 🪶 Lightweight venv | Chat, routing, dev, Python-only skills | pip install -e ".[interactive]" |
| 🖥️ Desktop/web backend | OmicsClaw-App or browser frontends | oc desktop-server --host 127.0.0.1 --port 8765 |
| 🧠 Memory API | Inspect graph memory over HTTP | pip install -e ".[memory]" then oc memory-server |
📖 Details: installation guide, quickstart. Dependencies live in pyproject.toml, environment.yml, and 0_setup_env.sh.
oc list and skills/catalog.json currently agree on 89 registered skills.
| Domain | Examples | Docs |
|---|---|---|
| 🧫 Spatial transcriptomics | QC, domains, annotation, deconvolution, CNV, trajectory | spatial |
| 🔬 Single-cell omics | QC, clustering, annotation, doublets, velocity, GRN | singlecell |
| 🧬 Genomics | QC, alignment, variants, CNV, assembly, epigenomics | genomics |
| 🧪 Proteomics | DIA/DDA, PTM, networks, biomarkers | proteomics |
| ⚗️ Metabolomics | Peaks, normalization, annotation, pathways | metabolomics |
| 📈 Bulk RNA-seq | DE, enrichment, co-expression, deconvolution, survival | bulkrna |
| 🧠 Orchestration | Routing, planning, literature support | orchestrator |
Run oc list for the current CLI catalog.
Graph-backed memory at omicsclaw/memory/ carries your sessions, datasets, analyses, preferences, and insights across runs — chat history and lineage come back when you reopen any surface. Each surface stays isolated so state never leaks across users or workspaces:
| Surface | Memory scope |
|---|---|
| CLI / TUI | Per workspace path |
| Desktop app | Per launch (or per signed-in user) |
| Telegram / Feishu bot | Per platform user |
A reserved __shared__ pool (core agent identity, knowledge handbook guards, glossary) is the one thing every surface reads back automatically. Full vocabulary and architecture in docs/CONTEXT.md.
Does OmicsClaw upload my raw data?
No. Skills run in the configured local or remote runtime; LLM calls should receive context and tool results, not raw omics matrices.
Which installation path should I use?
Use bash 0_setup_env.sh for real analysis. Use the lightweight venv only for chat, routing, development, or Python-only skills.
Can the desktop App run jobs on a server?
Yes. Run oc desktop-server on the remote Linux host, keep it bound to 127.0.0.1, and connect through the App's SSH tunnel runtime.
| Rule | Meaning |
|---|---|
| 🔒 Local-first | Raw data processing happens in your local or remote runtime |
| 🧪 Research use only | Not a medical device; no clinical diagnosis |
| 👩🔬 Expert review | Validate scientific outputs before decisions |
| 🔐 Remote caution | Use localhost binding, SSH tunnels, and tokens |
See data privacy and rules/disclaimer.
Maintainers: Luyi Tian (Principal Investigator), Weige Zhou (Lead Developer), Liying Chen (Developer), and Pengfei Yin (Developer).
🐛 Issues · 💬 Discussions · 📖 Docs
Architecture, skill design, and local-first philosophy are inspired by ClawBio, the first bioinformatics-native AI agent skill library. Memory and session-continuity patterns are inspired by Nocturne Memory.
- New skills: see CONTRIBUTING.md and the v2 scaffold under
templates/skill/. - Repository / agent work: see AGENTS.md — covers contract tests, provider contracts, skill runner, and architecture references.
Apache-2.0. See LICENSE.
@software{omicsclaw2026,
title = {OmicsClaw: A Memory-Enabled AI Agent for Multi-Omics Analysis},
author = {Zhou, Weige and Chen, Liying and Yin, Pengfei and Tian, Luyi},
year = {2026},
url = {https://github.com/TianGzlab/OmicsClaw}
}