User Guide Overview¶

Welcome to the Octopi User Guide! This comprehensive tutorial series will take you from raw tomogram data to precise particle coordinates using deep learning-based 3D particle picking.

Tutorial Sections¶

🏷️ Prepare Labels¶

Create training targets from particle annotations

Learn how to convert particle coordinates into 3D training masks. This section covers:

• Automated target generation from data portal annotations
• Manual specification for custom datasets
• Quality control and validation techniques
• Multi-class segmentation preparation

When to use: Start here after importing your data to prepare training materials.

🧠 Training¶

Train 3D U-Net models with Bayesian optimization

Master both single model training and automated architecture exploration:

• Model exploration with Bayesian optimization (recommended)
• Single model training for specific use cases
• MLflow experiment tracking and monitoring
• Best practices for resource management

When to use: After preparing training labels, use this to develop your particle picking models.

🔮 Inference and Localization¶

Apply trained models to generate predictions and extract particle coordinates

Deploy your trained models to analyze new tomograms: - Segmentation mask generation - Peak detection and particle extraction - Performance evaluation against ground truth

When to use: Once you have trained models, use this to get final particle coordinates.

What's Next?¶

Ready to start? Choose your entry point:

🚀 New to OCTOPI?¶

Follow the complete workflow:

• Begin with Labels → - Start the complete workflow
• Jump to Training → - If you already have training targets

🔬 Have existing models?¶

• Skip to Inference → - If you have pre-trained models

💻 Python developer?¶

• Explore the API → - For programmatic usage

Each tutorial builds on the previous ones, but you can jump to specific sections based on your needs and existing progress.