TrackMyPDB - Streamlit Application

A comprehensive bioinformatics pipeline for extracting heteroatoms from protein structures and finding molecularly similar compounds using fingerprint-based similarity analysis.

© 2025 Standard Seed Corporation. This is an open-source project developed and released by Standard Seed Corporation under the MIT License. All rights reserved.

🎯 Overview

TrackMyPDB is a user-friendly Streamlit web application that combines two powerful components:

1. Heteroatom Extraction Tool: Systematically extracts all heteroatoms from PDB structures associated with UniProt proteins
2. Molecular Similarity Analyzer: Finds ligands most similar to a target molecule using Morgan fingerprints and Tanimoto similarity

🚀 Quick Start

Web Application

• https://trackmypdbsscai.streamlit.app/

Launch using local machine

Prerequisites

• Python 3.7+
• Internet connection for API calls
• Windows OS (optimized for Windows environment)

Installation

1. Clone the repository:

   git clone <repository-url>
   cd TrackMyPDB

2. Install dependencies:

   pip install -r requirements.txt

3. Launch the application:

   streamlit run streamlit_app.py

4. Open your browser to http://localhost:8501

Basic Usage

1. Navigate to the web interface
2. Choose analysis type:
   • 🔍 Heteroatom Extraction
   • 🧪 Similarity Analysis
   • 📊 Complete Pipeline
3. Input your data:
   • UniProt IDs (e.g., Q9UNQ0, P37231, P06276)
   • Target SMILES structure
4. Run analysis and download CSV results

📋 Application Features

🔍 Heteroatom Extraction

• Input: UniProt protein identifiers
• Process: Fetches PDB structures, extracts heteroatoms, retrieves SMILES
• Output: Comprehensive CSV with chemical information
• APIs: RCSB PDB, PubChem integration
• Features: Progress tracking, error handling, result caching

🧪 Molecular Similarity Analysis

• Input: Target SMILES structure
• Process: Morgan fingerprint computation, Tanimoto similarity calculation
• Output: Ranked similarity results with interactive visualizations
• Features: Configurable parameters, real-time analysis, comprehensive reports

📊 Complete Pipeline

• Workflow: End-to-end processing from UniProt IDs to similarity results
• Integration: Automatic heteroatom extraction followed by similarity analysis
• Output: Both heteroatom database and similarity results

🏗️ Project Structure

TrackMyPDB/
├── streamlit_app.py              # Main Streamlit application
├── requirements.txt              # Python dependencies
├── backend/
│   ├── __init__.py              # Package initialization
│   ├── heteroatom_extractor.py  # Heteroatom extraction logic
│   └── similarity_analyzer.py   # Similarity analysis logic
└── README.md                    # This file

🛠️ Technical Details

Dependencies

• Streamlit: Web application framework
• RDKit: Cheminformatics and molecular similarity
• Pandas: Data manipulation and analysis
• Plotly: Interactive visualizations
• Requests: API communications
• NumPy: Numerical computations

API Integration

• PDBe REST API: PDB structure mappings
• RCSB PDB API: Chemical component data
• PubChem API: Backup molecular data

Molecular Analysis

• Morgan Fingerprints: Circular molecular fingerprints (radius=2, 2048 bits)
• Tanimoto Similarity: Industry-standard similarity metric (0-1 scale)
• Interactive Visualizations: Distribution plots, similarity rankings, statistical analysis

🎮 User Interface

Apple-Inspired Design

• Modern UI: Clean, minimalist design inspired by Apple Design principles
• Responsive Layout: Optimized for different screen sizes
• Interactive Elements: Smooth animations and hover effects
• Intuitive Navigation: Clear section organization and progress indicators

Key Features

• Real-time Progress: Progress bars and status updates
• Error Handling: Graceful error messages and troubleshooting
• Data Export: CSV download functionality with timestamps
• Result Caching: Session state management for efficiency

📊 Expected Results

Typical Output

• Heteroatoms: ~1000-5000 heteroatoms per 10 UniProt proteins
• SMILES Success: ~60-80% success rate for SMILES retrieval
• Similar Ligands: ~50-200 similar compounds per target (similarity > 0.2)
• Processing Time: 30-60 minutes for complete pipeline

File Outputs

• heteroatom_results_YYYYMMDD_HHMMSS.csv: Complete heteroatom extraction results
• similarity_results_YYYYMMDD_HHMMSS.csv: Molecular similarity analysis results

🔧 Configuration Options

Heteroatom Extraction

• UniProt IDs: Multiple input formats (comma-separated, line-separated)
• Result Caching: Previous results loading and management
• API Settings: Automatic retry logic and rate limiting

Similarity Analysis

• Fingerprint Parameters:
  • Morgan radius: 1, 2, 3 (default: 2)
  • Fingerprint bits: 1024, 2048, 4096 (default: 2048)
• Analysis Parameters:
  • Top N results: 10-100 (default: 50)
  • Minimum similarity: 0.0-1.0 (default: 0.2)

🚨 Troubleshooting

Common Issues

"Module not found" errors

# Install dependencies
pip install -r requirements.txt

# For RDKit installation issues on Windows
conda install -c conda-forge rdkit

"Invalid SMILES" warnings

• Verify SMILES syntax using online validators
• Check for special characters or formatting issues
• Example valid SMILES: CCO (ethanol), CC(=O)O (acetic acid)

Slow performance

• Reduce number of UniProt IDs for testing
• Use higher minimum similarity threshold
• Check internet connection stability

API timeout errors

• Wait a few minutes and retry
• Check if external APIs (RCSB, PubChem) are accessible
• Reduce batch size for large datasets

💡 Use Cases

Drug Discovery

• Lead Optimization: Find similar compounds to known drugs
• Scaffold Hopping: Identify alternative molecular frameworks
• Target Analysis: Understand ligand binding preferences

Chemical Biology

• Cofactor Analysis: Study enzyme cofactor preferences
• Binding Site Analysis: Characterize pocket properties
• Cross-reactivity Prediction: Assess off-target binding

Academic Research

• Structural Biology: Build custom screening libraries
• Comparative Analysis: Study protein-ligand interactions
• Database Construction: Create specialized molecular databases

🤝 Contributing

Development Guidelines

• Follow PEP 8 style guidelines
• Add comprehensive error handling
• Include progress indicators for long operations
• Document all functions and classes
• Test with various input formats

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

Open Source Project - Free to use, modify, and distribute under the MIT License terms.

Please respect API terms of service and rate limits when using this application.

🙏 Acknowledgments

• RCSB PDB: Protein structure data
• PDBe: Structure mapping services
• PubChem: Chemical information database
• RDKit: Cheminformatics toolkit
• Streamlit: Web application framework

---

👨‍💻 Developers

• Prject Lead/Senior Engineer: Sul sharif
• Lead Engineer: Anu Gamage
• Associate Engineers: Damilola Bodun, Kalana Kotawalagedara & Logan Geffen

📞 Support

For issues or questions:

1. Check the troubleshooting section
2. Verify input data format
3. Test with provided examples
4. Review browser console for errors
5. Contact the developers through LinkedIn

Happy molecular hunting! 🧬🔍