AlignAIR

AlignAIR is an AGPL-3 licensed sequence alignment tool specifically designed for Adaptive Immune Receptor Repertoire (AIRR) sequences. AlignAIR v2.0 features a unified architecture that dynamically supports both single-chain and multi-chain analysis with seamless GenAIRR integration.

Overview

AlignAIR v2.0 represents a major architectural advancement, combining powerful sequence alignment algorithms with a unified, modular interface. The new system features:

• Unified Model Architecture: SingleChainAlignAIR and MultiChainAlignAIR models that automatically adapt to different receptor types
• Dynamic GenAIRR Integration: Built-in support for GenAIRR dataconfigs with automatic chain type detection
• Multi-Chain Analysis: Native support for analyzing mixed receptor populations (e.g., IGK + IGL light chains)
• Flexible Data Handling: Universal dataset classes that work with any combination of chain types

Features

• Unified Multi-Chain Framework: AlignAIR v2.0 introduces a revolutionary unified architecture that seamlessly handles single-chain and multi-chain scenarios
• Dynamic GenAIRR Integration: Built-in support for GenAIRR dataconfigs with automatic chain type detection and configuration
• Universal Model Architecture: SingleChainAlignAIR and MultiChainAlignAIR models that adapt to any receptor combination
• Enhanced Multi-Task Learning: Joint optimization of V/D/J segmentation, allele calling, chain type classification, and sequence analysis
• Scalable Performance: Optimized for large-scale AIRR data with efficient batch processing and GPU acceleration
• Comprehensive Chain Support: Native support for IGH, IGK, IGL, TCRB, and custom receptor types

Installation Options

To install the latest stable release of AlignAIR, use:

pip install AlignAIR

For installation from the GitHub repository for the latest development version:

git clone https://github.com/MuteJester/AlignAIR.git
cd AlignAIR
pip install .

Quick Start Guide

1. Input Preparation: Ensure your input data is in a compatible format (e.g., FASTA or CSV with sequences).
2. Choose Configuration: Select appropriate GenAIRR dataconfig(s) for your receptor type(s):
   • Single chain: --genairr-dataconfig=HUMAN_IGH_OGRDB
   • Multi-chain: --genairr-dataconfig=HUMAN_IGK_OGRDB,HUMAN_IGL_OGRDB
3. Running AlignAIR: Use the unified CLI interface:

    python app.py run --model-dir=model_path \
                     --genairr-dataconfig=HUMAN_IGH_OGRDB \
                     --sequences=my_sequences.csv \
                     --save-path=results/
   

4. Results: AlignAIR automatically detects single vs. multi-chain scenarios and adapts accordingly.

Docker Support

AlignAIR provides a Docker image to ensure a consistent runtime environment. To use AlignAIR with Docker:

1. Pull the Docker Image:

    docker pull thomask90/alignair:latest
   

2. Run the Container (entrypoint style): bash docker run -it --rm \ -v $(pwd):/data \ -v $(pwd)/results:/downloads \ thomask90/alignair:latest run \ --model-dir=/app/pretrained_models/IGH_S5F_576 \ --genairr-dataconfig=HUMAN_IGH_OGRDB \ --sequences=/data/my_sequences.fasta \ --save-path=/downloads/

Documentation Resources

• Getting Started Guide – Learn how to set up and start using AlignAIR.
• Configuration Reference – Detailed explanations of all customizable parameters.
• CLI Commands – A comprehensive list of commands and options.
• Examples and Tutorials – Hands-on examples to help users familiarize themselves with AlignAIR’s capabilities.

Development and Contributions

AlignAIR is an open-source project. We welcome contributions from the community:

• GitHub Repository
• Contributing Guide
• Issue Tracker – Report bugs or suggest new features.

Publications and References

The detailed methodology and performance benchmarks are discussed in the main manuscript and supplementary documentation here.

This site is open source. Improve this page.