🧬 Biopharma M&A Radar

A real-time web application for tracking biotech and biopharma mergers, acquisitions, and partnerships using the GDELT 2.0 DOC API.

🚀 Quick Start

1. Setup & Installation

# Navigate to project directory
cd '/Users/junchilu/Desktop/2025/AI & Data Science/00_Projects/easy_prompt_app'

# Create virtual environment (if not exists)
python3 -m venv venv

# Activate virtual environment
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

2. Run the App

# Option 1: Direct command
streamlit run app.py

# Option 2: Use startup script
./start_app.sh

3. Access the App

• Local URL: http://localhost:8501
• Network URL: http://192.168.0.13:8501

📁 Project Structure

easy_prompt_app/
├── app.py                    # 🎯 MAIN APPLICATION (635 lines)
├── requirements.txt          # 📦 Python dependencies
├── start_app.sh             # 🚀 Quick launcher script
├── venv/                    # 🐍 Virtual environment
└── README.md                # 📖 This comprehensive guide

Single Source of Truth: Everything is in app.py - no confusion about which file to run!

🎯 Features

Real-time Data

• GDELT 2.0 DOC API Integration: Fetches biotech news from thousands of sources
• No API Key Required: Free public API access
• Smart Filtering: Automatically filters for biotech/pharma deals

Intelligent Data Extraction

• Deal Size Parsing: Converts "$2.3B", "800 million", "1.2bn" to USD millions
• Company Extraction: Identifies acquirer and target companies
• Deal Type Classification: Acquisition, merger, partnership, investment
• Deduplication: Removes duplicate articles by URL

Interactive Visualizations

• KPI Dashboard: Total deals, deals with size, median/largest deal values
• Bar Chart: Top 10 deals by size (Acquirer → Target)
• Timeline: Deal dots over time with size proportional to value
• Network Graph: Acquirer-target relationships with edge thickness = deal size
• Data Table: Complete deal information with clickable source links

🛠️ Tech Stack

• Backend: Python 3.13.5
• Frontend: Streamlit
• Data Processing: Pandas, NumPy
• Visualizations: Plotly, NetworkX, Matplotlib, Seaborn
• API: GDELT 2.0 DOC API

📊 How to Use

1. Launch the App

source venv/bin/activate
streamlit run app.py

2. Configure Search

• Use the sidebar to set search parameters
• Adjust date range (30-365 days)
• Customize search queries

3. Fetch Data

• Click "Fetch Deals Data" to search for recent deals
• The app automatically searches multiple biotech-related queries
• Results are filtered for biotech/pharma relevance

4. Explore Results

• View KPI cards with key metrics
• Analyze top deals in the bar chart
• Track deal timeline
• Explore company relationships in network graph
• Browse detailed deals table

🔍 Data Sources & Processing

GDELT API Queries

The app automatically searches for:

• "biotech acquisition AND (merck OR abbvie OR pfizer OR novartis OR roche)"
• "biopharma merger AND (therapeutics OR pharmaceuticals OR biotech)"
• "pharmaceutical acquisition AND (drug OR medicine OR therapy)"
• "biotech partnership AND (drug development OR clinical trial)"
• "pharma merger AND (biotech OR therapeutics)"
• "biotech buyout AND (drug OR medicine)"

Data Extraction

• Deal Size Patterns: Recognizes $2.3B, 800 million, 1.2bn USD, etc.
• Company Names: Extracts acquirer and target from headlines
• Deal Types: Classifies as acquisition, merger, partnership, investment
• Biotech Filtering: Keeps only biotech/pharma related deals

📈 Sample Results

The app successfully extracts deals like:

• Merck nears $10 billion deal for biotech Verona → $10,000M, Merck → Verona
• AbbVie Finalizes Acquisition Of Capstan Therapeutics → AbbVie → Capstan Therapeutics
• Recursion Pharmaceuticals acquiring REV102 rights → Recursion Pharmaceuticals → REV102

🔧 Configuration

Search Parameters

• Date Range: 30-365 days lookback
• Max Records: 250 per query (configurable)
• Search Terms: Customizable biotech queries

Deal Size Recognition

• $2.3B, $2.3 billion, 2.3B dollars
• $800M, $800 million, 800M USD
• $500K, $500 thousand, 500K dollars

🚧 Current Limitations

• Basic company name extraction (could use NER)
• Limited deal type classification
• No historical data persistence
• Basic deduplication logic

🔮 Future Enhancements

• Named Entity Recognition (NER) for better company extraction
• Machine learning for deal type classification
• Database integration for historical tracking
• Advanced filtering and search capabilities
• Email alerts for new deals
• Export to Excel/CSV functionality

🆘 Troubleshooting

If the app doesn't start:

# Check virtual environment
source venv/bin/activate
python --version

# Reinstall dependencies
pip install -r requirements.txt

If you get port conflicts:

# Use a different port
streamlit run app.py --server.port 8502

If API calls fail:

• Check your internet connection
• Verify GDELT API is accessible
• Check the Streamlit logs for error messages

📝 API Information

GDELT 2.0 DOC API

• Endpoint: https://api.gdeltproject.org/api/v2/doc/doc
• Authentication: None required
• Rate Limits: Reasonable usage limits apply
• Documentation: GDELT Project

🤝 Contributing

This is a workshop project designed for easy modification and extension. Key areas for improvement:

1. Data Extraction: Enhance company name and deal size parsing
2. Visualizations: Add more chart types and interactive features
3. Data Storage: Implement database persistence
4. UI/UX: Improve the Streamlit interface
5. Performance: Optimize API calls and data processing

📄 License

This project is created for educational and workshop purposes.

---

🎨 Vibe Coding Prompt

Here's the effective prompt to create this Biopharma M&A Radar app:

🚀 Cursor Kick-off Prompt

I want to build a Biopharma M&A Radar web app.

Requirements:

Stack: Use Python + Streamlit (preferred for fast prototyping).

API source: Use the GDELT 2.0 DOC API (JSON, no API key required) to fetch recent news headlines about biotech / biopharma mergers, acquisitions, and partnerships.

Data extraction: For each article, parse the headline, URL, date, acquirer, target, deal size (USD), and deal type. Use regex and heuristics to capture deal size (e.g., $2.3B, 800 million, etc.) and normalize to USD numeric values.

Deduplicate by URL or similar titles. Filter only healthcare/biopharma deals.

Visualization:

KPI cards (total deals, # with deal size, median deal size, largest deal).

Bar chart: Top 10 deals by size (Acquirer → Target, Year).

Timeline: Deal dots over time, size = deal value.

Breakdown by big pharma companies: Pie chart and bar chart showing deal distribution.

Table: Show extracted fields in a clean table with links to the source articles.

Keep code clean, modular, and well-commented so I can edit easily in the workshop.

Deliverables in this session:

Create a Streamlit app skeleton with placeholders for data fetching, parsing, and visualizations.

Implement the GDELT API fetch function with a test query for "biotech acquisition" in the last 6 months.

Return raw JSON results in a Streamlit preview to confirm API works.

Key considerations:
- Handle timezone issues in date comparisons
- Improve company name extraction with known biotech/pharma company lists
- Add realistic sample deal sizes for demonstration
- Filter for recent deals only (within specified date range)
- Remove network graph visualization (too complex)
- Focus on big pharma breakdown instead
- Clean up folder structure - consolidate documentation
- Ensure single source of truth (one main app.py file)

---

🎉 Ready to Go!

Just run: streamlit run app.py

That's it! No confusion, no multiple files to choose from. Everything you need is in the main app.py file.

Built with ❤️ for the biopharma community