About AlignAIR
AlignAIR: Enhanced Sequence Alignment of Adaptive Immune Receptors Using Multi-Task Deep Supervised Learning
Why?
The analysis of Adaptive Immune Receptor Repertoire sequencing (AIRR-seq) data is critical for understanding the dynamics of the adaptive immune system in health and disease. While many algorithms and tools are available for Immunoglobulin (Ig) sequence alignment, each exhibits specific strengths and weaknesses. AlignAIR was developed to harness a novel multi-task residual convolutional architecture along with an immunology-informed loss function to overcome these weaknesses.
Comprehensive Sequence Analysis
AlignAIR integrates a deep residual convolutional architecture with a novel loss function and training regime, enabling effective modeling of Ig sequences. Using the immune receptor sequence simulator, GenAIRR, we generated high-quality training and validation datasets.
Superior Performance
AlignAIR was benchmarked against leading aligners, showing superior performance in segmentation tasks, allele classification, and productivity classification, especially under high Somatic Hypermutation (SHM) conditions. It accurately models SHM dynamics and resolves uncertainties by prioritizing alleles with higher mutability.
Robust Against Sequence Corruption
AlignAIR’s robust performance indicates its enhanced ability to deal with sequence corruptions and complex mutation patterns, which are often seen in clinical samples. It transcends classic alignment algorithms by integrating multiple metatasks into its latent space, allowing it to consider multiple aspects simultaneously when making an alignment.
Explore AlignAIR!
Gain access to the beta version of our web interface and explore the innovative features that make AlignAIR a game-changer in immunogenomics. Reach out to us with any questions or suggestions for development.