Quantification of Inter-Sample Differences in T-Cell Receptor Repertoires Using Sequence-Based Information.

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• Comparing T-cell receptor (TCR) repertoires is vital for understanding immune states across diverse biological contexts.
• Existing methods often rely on sequence overlap or abundance, which can be limited by data sparsity and fail to identify key sub-repertoires.
• These limitations hinder accurate assessment of inter-sample repertoire differences.

Purpose of the Study:

• To introduce RECOLD (REpertoire COmparison in Low Dimensions), a novel approach for quantitative comparison of TCR repertoires.
• To address the limitations of existing methods by utilizing sequence information for more robust comparisons.
• To identify specific TCR sequences contributing to inter-sample variations.

Main Methods:

• RECOLD embeds pairwise sequence dissimilarities into a low-dimensional space.
• Inter-sample repertoire differences are quantified using information-theoretic measures on estimated data distributions.
• The method leverages high-dimensional sequence information to overcome data sparsity.

Main Results:

• RECOLD accurately identifies hierarchical structures in mouse and human TCR repertoires, aligning with biological expectations.
• The method demonstrates robust performance on sparse TCR sequence datasets, outperforming previous approaches.
• RECOLD successfully identified specific TCR sequences driving differences in mouse repertoires with varying genetic backgrounds.

Conclusions:

• RECOLD provides a powerful and accurate tool for TCR repertoire comparison, especially in data-sparse scenarios.
• The method's ability to identify contributing sequences offers valuable insights for developing targeted immunotherapies and vaccines.
• This approach enhances our understanding of immune cell repertoire dynamics and variation.