8cubeDB: Mouse Specificity Explorer 🐭

Exploring gene specificity in 8 tissues from 8 individuals for each of the 8 founder mouse strains

8cubeDB provides a unified platform to explore gene specificity, marker genes, and expression variability across founder mouse tissues from the Rebboah et al. (2025) dataset. It includes both a RESTful API (built with FastAPI), an interactive dashboard (built with Streamlit) and MCP Server integration for LLM-powered data analysis.

---

🌐 Access the Platform

• Frontend Dashboard: https://mouseexplorer.onrender.com Explore genes, visualize Psi-blocks, and browse marker and housekeeping genes interactively.

• Backend API: https://eightcubedb.onrender.com/docs Programmatic access to the dataset via REST API. Refer to this colab notebook for a tutorial on how to query the database in python.

• MCP Server (Hosted SSE Endpoint): https://eightcubedb.onrender.com/mcp/sse Enables direct integration with LLM clients that support the Model Context Protocol (MCP).

---

System Overview

FastAPI Backend

The backend serves data from two curated SQLite databases:

• 8cube.db — primary specificity and Psi-block data
• mean_var_DB.db — mean and variance of gene expression across conditions

All API routes stream data as CSV downloads for seamless integration with downstream tools.

Key Endpoints

Endpoint	Description
/	API root — overview of available endpoints
/config	Returns available analysis levels, types, and block labels
/specificity	Extract gene specificity data for given genes
/psi_block	Fetch Psi-block data by analysis level/type
/highly_specific	Retrieve genes highly specific to a given variable
/non_specific	Retrieve non-specific (housekeeping) genes
/marker	Identify marker genes by block label
/gene_expression	Get gene expression mean and variance values

---

Streamlit Frontend

The Mouse Specificity Explorer dashboard provides an interactive interface to visualize and query the API.

Main Features

Tab	Functionality
🔬 Gene Viewer	Visualize Psi-block and gene expression data for specific genes
🗺️ Specificity Explorer	Browse gene specificity across the dataset
⭐ Highly Specific Genes	Identify genes specific to a tissue or condition
🏠 Housekeeping Genes	Explore broadly expressed, non-specific genes
🎯 Marker Genes	Discover marker genes for selected blocks

Built with Streamlit, Plotly, and Pandas, the app offers clean visualizations and downloadable tables.

---

🤖 AI & LLM Integration (MCP)

8cubeDB supports the Model Context Protocol (MCP), allowing AI assistants like Claude to query the database directly to answer questions about mouse gene specificity.

1. Hosted MCP (Remote)

You can connect your MCP-compatible client to our hosted SSE endpoint:

• Endpoint: https://eightcubedb.onrender.com/mcp/sse

2. Local MCP (For Claude Desktop)

To use 8cubeDB directly within the Claude Desktop app, follow these steps:

1. Download the mcp_server_local_for_claude.py file from this repository.
2. Locate your config file:

• macOS: ~/Library/Application\ Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

3. Add the following configuration (ensure you update the paths to your specific Python environment and the location of the downloaded script):

{
  "mcpServers": {
    "8cubedb": {
      "command": "/YOUR/PATH/TO/python",
      "args": [
        "/YOUR/PATH/TO/mcp_server_local_for_claude.py"
      ]
    }
  }
}

---

Data Summary

• Dataset: Rebboah et al. (2025) — 8cube founder mouse dataset
• Metrics: Ψ (Psi) specificity index and ζ (Zeta) selectivity metric
• Levels: Multi-scale analysis (cell type, tissue, organ, etc.)
• Sources: Derived from table_1 (global summary) and *_psi_block tables (block-level metrics)

---

⚙️ Architecture Overview

                        ┌──────────────────────────────┐
                        │        SQLite Databases      │
                        │   - 8cube.db                 │
                        │   - mean_var_DB.db           │
                        └───────────────┬──────────────┘
                                        │ SQLite Queries
                                        ▼
                        ┌──────────────────────────────┐
                        │        FastAPI Backend       │
                        │  (eightcubedb.onrender.com)  │
                        └───────────────┬──────────────┘
                                        │
                        ┌───────────────┴───────────────┐
                        │                               │
                        ▼                               ▼
        ┌──────────────────────────────┐   ┌──────────────────────────────┐
        │        Streamlit Frontend    │   │        MCP Server (SSE)      │
        │  mouseexplorer.onrender.com  │   │  /mcp/sse or local Claude    │
        └──────────────────────────────┘   └──────────────────────────────┘

---

Project Attribution

Developed by Nikhila P. Swarna Pachter Lab, Caltech

Part of the IGVF Consortium Specificity analyses powered by ember.

---

License

Licensed under the BSD 2-Clause License. © 2025 Pachter Lab · All rights reserved.

---

Citation

If you use this website, MCP server or the accompanying database, please cite the following papers:

• Rebboah E, et al. Systematic cell-type resolved transcriptomes of 8 tissues in 8 lab and wild-derived mouse strains captures global and local expression variation. (2025) DOI: https://doi.org/10.1101/2025.04.21.649844

• Swarna NP, et al. Determining gene specificity from multivariate single-cell RNA sequencing data. (2025) DOI: https://doi.org/10.1101/2025.11.21.689845