A spectral clustering-based method for identifying clones from high-throughput B cell repertoire sequencing data

Area of Science:

• Immunology
• Computational Biology
• Bioinformatics

Background:

• B cells generate diverse antigen receptors (Immunoglobulin, Ig) through DNA rearrangement and somatic hypermutation.
• Next-generation sequencing allows large-scale profiling of B cell Ig repertoires.
• Accurate partitioning of Ig sequences into clonal groups is a key computational challenge.

Purpose of the Study:

• To develop a novel, computationally efficient method for partitioning B cell Ig sequences into clones.
• To improve the accuracy and sensitivity of clonal identification in repertoire sequencing data.

Main Methods:

• Proposed a new method utilizing spectral clustering with an adaptive threshold for sequence neighborhood determination.
• Validated the method using simulated and experimental B cell repertoire datasets.
• Compared performance against existing fixed-threshold and likelihood-based methods.

Main Results:

• The spectral clustering method demonstrated high sensitivity and specificity in identifying B cell clones.
• Outperformed fixed-threshold methods, especially in datasets where optimal threshold selection is difficult.
• Achieved greater computational efficiency compared to existing methods.

Conclusions:

• The proposed spectral clustering approach provides a robust and efficient solution for B cell clonal partitioning.
• Enables more accurate downstream analyses, including diversity metrics, lineage reconstruction, and selection analysis.
• Represents a significant advancement for B cell repertoire analysis.