lobster-local

内容来源:README.md(说明文档) · 原始地址 · 查看安装指南

原始内容

🦞 Lobster

License: AGPL-3.0-or-later Documentation: CC BY 4.0 Python 3.11+ PyPI

Bioinformatics co-pilot to automate redundant omics-related tasks so you can focus on science


📋 Table of Contents


✨ What is Lobster AI?

Lobster AI combines specialized AI agents with open-source bioinformatics tools to analyze multi-omics data through natural language. Describe your analysis needs - no coding required.

Perfect for:

  • 🧬 Bioinformatics researchers analyzing RNA-seq data
  • 💻 Computational biologists seeking intelligent workflows
  • 🔬 Life science teams needing reproducible results
  • 📚 Students learning modern bioinformatics

⚡ Quick Start

# Install Lobster globally
uv pip install lobster-ai

# Configure (interactive wizard)
lobster init

# Start analyzing
lobster chat

Don't have uv? Install it: macOS/Linux | Windows

Alternative installation methods: See Complete Installation Guide or Wiki


📺 Getting Started

Installation

1️⃣ Install in 30 seconds
uv pip install lobster-ai

Configuration

2️⃣ Configure in 1 minute
lobster init

Chat Demo

3️⃣ Start analyzing with natural language


🏗️ Architecture Overview

Lobster is a modular bioinformatics platform with pluggable execution environments, LLM providers, and integrated data management:

Lobster Architecture

Component Matrix

Layer Component Configuration Use Case
Execution Local Default (no setup) Privacy-first, offline, cost-sensitive
Cloud LOBSTER_CLOUD_KEY Team collaboration, scaling, managed infrastructure
LLM Provider Ollama ollama pull gpt-oss:20b Local-only, unlimited usage, offline
Anthropic ANTHROPIC_API_KEY Best quality, quick start, cloud/local
AWS Bedrock AWS credentials Enterprise, compliance, high throughput
Google Gemini GOOGLE_API_KEY Long context, multimodal, free tier available
Data Sources GEO/SRA/ENA Auto-configured Transcriptomics datasets
PRIDE/MassIVE Auto-configured Proteomics datasets
PubMed/PMC NCBI_API_KEY (optional) Literature mining, metadata extraction
Data Management DataManagerV2 Auto-configured Multi-modal data orchestration, provenance tracking

Configuration & Deployment

Lobster supports four deployment patterns optimized for different use cases:

Pattern Best For Key Features
Local + Ollama Privacy, learning, zero cost Offline, unlimited usage, 100% local
Local + Anthropic Quality, development Best accuracy, quick setup, flexible
Local + Gemini Cost optimization, free tier Long context, multimodal, free tier available
Cloud + Bedrock Production, teams Enterprise SLA, high limits, scalable

Learn more:


🧬 Features

Category Capabilities Docs Status
🧬 Single-Cell RNA-seq QC, Clustering, Annotation, Markers, Trajectories Tutorial ✅ Available
📊 Bulk RNA-seq Differential expression, Complex designs, Formula-based stats Tutorial ✅ Available
📚 Literature Mining PubMed search, Full-text retrieval, Methods extraction Guide ✅ Available
🗄️ Dataset Discovery GEO, SRA, PRIDE, ENA search and validation Guide ✅ Available
📈 Visualization UMAP, Volcano plots, Heatmaps, Interactive figures Examples ✅ Available
🧪 Metadata Assistant Cross-dataset harmonization, ID mapping Guide ⭐ Premium
🔬 Proteomics DDA/DIA workflows, Missing values, Normalization Tutorial ⭐ Premium

Full documentation →


📊 Interactive Dashboard (Alpha)

Dashboard Demo

Real-time analysis monitoring with live updates and visualization

Dashboard for deep-diving into omics anlaysis.

#start with 
lobster dashboard
#or during CLI session with
❯ /dashboard

📘 Case Studies

🧬 Single-Cell Analysis

Download → QC → Cluster → Annotate in one conversation

❯ Download GSE109564, perform QC, cluster cells, and find markers

✓ Downloaded 5,000 cells × 20,000 genes
✓ Quality control: filtered to 4,477 high-quality cells
✓ Identified 12 distinct clusters
✓ Generated UMAP visualization with marker genes

📚 Literature Mining

Find papers, extract methods, discover datasets

❯ Find papers about CRISPR screens in cancer and extract their GEO datasets

Found 47 papers with 23 associated GEO datasets
Extracted methods from 12 papers with full-text access
Cached metadata for all datasets in workspace

📈 Visualization

Publication-ready figures with natural language

❯ Create a UMAP colored by cell type with cluster labels

Generated interactive UMAP visualization
Saved as: geo_gse109564_umap_celltype.html
Also exported as PNG for publications

More examples in our cookbook →


🗓️ Roadmap

2026 Development:

  • Custom feature agent (bring your tools/best practices)
  • Knowledge graph integration for multi-dataset analysis
  • Lobster Cloud compute infrastructure
  • Enhanced multi-omics workflows (MuData integration)
  • Community-contributed agent marketplace

Submit feature ideas: GitHub Discussions


🔧 Use with Claude Code

Lobster integrates with Claude Code as an Agent Skill, allowing you to run bioinformatics analyses directly from your development environment.

Quick Setup

# 1. Install Lobster
uv pip install lobster-ai
lobster init

# 2. Install Lobster Skill for Claude Code
curl -fsSL https://raw.githubusercontent.com/the-omics-os/lobster-local/main/claude-skill/SKILL.md \
  -o ~/.claude/skills/lobster/SKILL.md --create-dirs

# 3. Start Claude Code
claude

Usage Examples

Claude Code will automatically invoke Lobster when you mention bioinformatics tasks:

# In Claude Code, just ask naturally:
You: "Analyze the single-cell dataset in data/counts.h5ad"
Claude: I'll use Lobster to analyze this single-cell dataset...
        [Invokes: lobster query "analyze data/counts.h5ad..."]

You: "Download GSE109564 and run quality control"
Claude: I'll use Lobster to download and analyze this dataset...
        [Invokes: lobster query "download GSE109564 and run QC"]

You: "Find papers about CRISPR screens in cancer"
Claude: I'll use Lobster to search the literature...
        [Invokes: lobster query "search pubmed for CRISPR screens cancer"]

What this enables:

  • 🧬 Run bioinformatics analyses without leaving your IDE
  • 📊 Extract data and generate reports programmatically
  • 🔄 Combine Lobster with your development workflow
  • 🤖 Let Claude handle the complexity of bioinformatics commands

Learn more about Agent Skills →


📚 Documentation

Resource Description
Getting Started First-time user guide
Installation Detailed installation options
Configuration LLM providers, API keys, settings
CLI Commands Interactive command reference
Examples Cookbook Real-world analysis recipes
Troubleshooting Common issues and solutions
FAQ Frequently asked questions

Full reference: README_FULL.md | Wiki Home


🤝 Community & Support

Contributing: We welcome contributions! See CONTRIBUTING.md for guidelines.


📄 License

Lobster AI is open source under AGPL-3.0-or-later.

This license ensures all users receive the freedoms to use, study, share, and modify the software. Documentation is licensed CC-BY-4.0.

Commercial licensing: Contact info@omics-os.com


Transform Your Bioinformatics Research Today

uv pip install lobster-ai && lobster chat

Made with sleep deprivation ☕️ by Omics-OS