LipiDetective

Deep learning for lipid identification from tandem mass spectra.

LipiDetective is a transformer-based framework that identifies molecular lipid species from MS/MS spectra. Given a tandem mass spectrum, it predicts the molecular lipid species name in shorthand notation — including headgroup and fatty acid composition.

Note

This project is under active development.

Getting Started

1. Install LipiDetective

2. Follow the Quickstart to run your first prediction

3. See the Configuration guide for all available options

User Guide

• Installation
  • Requirements
  • Install with UV (Recommended)
  • Install from Source with pip
  • Development Setup
  • Verify Installation
• Quickstart
  • 1. Copy the Config Template
  • 2. Edit the Configuration
  • 3. Run the Prediction
  • 4. View the Results
  • Next Steps
• Configuration
  • Model Selection
  • CUDA
  • File Paths
  • Workflow
  • Training
  • Test
  • Prediction
  • Hyperparameter Tuning
  • Input Embedding
  • Transformer
  • WandB Integration
  • Comment
• Data Preparation
  • Supported Formats
  • Converting mzML to HDF5
  • Spectrum Processing
• Workflows
  • Training
  • Validation
  • Testing
  • Prediction
  • Hyperparameter Tuning
  • Trainer API

Reference

• Models
  • Transformer (Recommended)
  • Convolutional Neural Network
  • Feed-Forward Network
  • Random Forest
• Output & Metrics
  • Training Output
  • Testing Output
  • Prediction Output
  • Tuning Output
  • WandB Integration
  • Custom Evaluation
• API Reference
  • Models
  • Workflow
  • Helpers

Indices and Tables

• Index

• Module Index