A multi-objective based clustering for inferring BCR clonal lineages from high-throughput B cell repertoire data

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• B cell adaptive responses involve clonal lineage expansion, mutation, and selection.
• Identifying B cell clonal lineages is crucial for repertoire analysis, tracking, and statistical studies.
• Existing clustering methods for B cell repertoire data typically optimize a single objective, potentially limiting performance.

Purpose of the Study:

• To introduce MobiLLe, a novel method for grouping clonally related B cell sequences using multi-objective clustering.
• To evaluate MobiLLe's performance against existing tools on simulated and experimental high-throughput sequencing data.
• To provide an efficient and accurate tool for B cell repertoire analysis, particularly in clinical settings.

Main Methods:

• MobiLLe utilizes V(D)J annotations for initial grouping.
• The method iteratively applies two objective functions to simultaneously optimize cohesion and separation within clonal lineages.
• Performance is assessed on simulated datasets with varying mutation rates and experimental repertoire data.

Main Results:

• MobiLLe demonstrates improved clonal lineage grouping on simulated benchmarks compared to other tools.
• Clustering results on experimental repertoires are comparable to leading methods and consistent with previous publications.
• MobiLLe achieves the lowest running time among state-of-the-art tools for repertoire analysis.

Conclusions:

• MobiLLe accurately identifies clonally related antibody sequences using multi-objective clustering.
• The method offers a significant improvement in efficiency and accuracy for B cell repertoire analysis.
• MobiLLe has the potential to advance the understanding of B cell malignancies and their evolution.