omicverse/omicverse

OmicVerse – a unified Python toolkit for modern transcriptomics & multi‑omics

What it is – OmicVerse is a Python library that brings together the whole workflow for bulk RNA‑seq, single‑cell RNA‑seq, spatial transcriptomics and related multi‑omics data. It bundles dozens of well‑known analysis tools (Scanpy, scVI, MOFA, CellPhoneDB, etc.) and adds a thin AI‑assisted layer (J.A.R.V.I.S. agents, Model‑Context‑Protocol server) so you can drive the same pipelines with natural‑language commands or integrate them into larger AI systems.

Why it matters – Researchers often have to stitch together many separate packages, each with its own data format and API. OmicVerse standardises on the anndata/mudata data structures, provides a single installable distribution (via PyPI, conda‑forge or Docker), and ships a curated collection of state‑of‑the‑art methods for preprocessing, integration, trajectory inference, cell‑type annotation, spatial deconvolution, and even downstream synthetic‑biology modelling. The added agent interfaces let you ask a language model to run a step (e.g., “annotate the lung scRNA‑seq”) without writing code.

---

Core capabilities

Area	What OmicVerse offers
Data backbone	Unified handling of pandas, numpy, anndata, mudata objects.
Bulk & single‑cell pipelines	Normalisation, differential expression, clustering, trajectory analysis, batch correction (Harmony, Scanorama, Combat), factor analysis (MOFA, scVI).
Spatial transcriptomics	Tools like STAligner, STAGATE, GraphST, and spatial deconvolution (Bulk2Space).
Cell‑cell communication	CellPhoneDB, COSG, etc.
AI‑assisted workflows	J.A.R.V.I.S. command‑line agents (omicverse claw …), MCP server exposing tools to LLMs (Claude, etc.), Telegram bot interface.
Synthetic biology extension (ov.synbio)	Metabolic‑model integration (COBRApy, GECKO), enzyme‑design predictions (ESMFold, ProteinMPNN, ESM‑2 embeddings), DNA design utilities.
Molecular dynamics extension (ov.mol)	Pull structures from AlphaFold/ESMFold, dock with AutoDock Vina, run GPU‑accelerated MD with OpenMM, compute MM‑GBSA binding energies.
Visualization & reporting	Helpers for complex heatmaps, interactive plots, and a web‑based analysis front‑end (omicverse_web).

---

Getting started

# Core package
conda install -c conda-forge omicverse   # or pip install -U omicverse

# Optional AI‑agent support
pip install "omicverse[jarvis]"

# Optional synthetic‑biology stack
pip install "omicverse[synbio]"

# Optional molecular‑dynamics stack
pip install "omicverse[md]"

All installations assume you have a compatible PyTorch build already installed.

Documentation & tutorials

• Read‑the‑Docs: https://omicverse.readthedocs.io/en/latest/
• Installation guide: https://starlitnightly.github.io/omicverse/Installation_guild/
• LLM / agent tutorials:
  • omicverse claw – natural‑language annotation (e.g., omicverse claw 'help me annotate the lung scrna‑seq').
  • MCP server – python -m omicverse.mcp to expose tools to AI assistants.
  • Telegram bot – omicverse jarvis --channel telegram ….
• Domain‑specific extensions: tutorials for ov.synbio and ov.mol are linked from the docs.

---

Who should use it?

• Computational biologists who need a single, well‑maintained environment for bulk, single‑cell, and spatial RNA‑seq analyses.
• Bio‑informaticians looking to integrate cutting‑edge AI models (LLMs, protein‑design nets) into their pipelines without writing glue code.
• Synthetic‑biology engineers interested in linking enzyme‑design predictions to genome‑scale metabolic models.
• Researchers who prefer a command‑line or chat‑based interface for routine analyses.

---

Ecosystem & community

• Part of the scverse ecosystem (the same umbrella as Scanpy, anndata, mudata, scvi‑tools).
• Distributed on PyPI, conda‑forge, and as a Docker image (starlitnightly/omicverse).
• Open‑source under the GNU GPL license; citations tracked via a DOI (Nature Communications 2024).
• Continuous integration via GitHub Actions; documentation automatically built on Read‑the‑Docs.

---

Bottom line: OmicVerse is a genuine, research‑grade software project that consolidates a massive collection of transcriptomics tools, adds AI‑driven agents for natural‑language control, and extends into synthetic biology and molecular dynamics—all from a single Python package.