Predicting class switch recombination in B-cells from antibody repertoire data

Area of Science:

• Immunology
• Computational Biology
• Bioinformatics

Background:

• Class switch recombination (CSR) is crucial for antibody response diversity.
• Predicting CSR is vital for understanding adaptive immunity.
• Existing methods lack robust predictive capabilities for CSR events.

Purpose of the Study:

• To develop and evaluate a machine learning framework for predicting CSR in B-cells.
• To analyze antibody repertoire data using clonal group (CG) features for CSR prediction.
• To compare the performance of various machine learning models in predicting CSR.

Main Methods:

• Utilized statistical and machine learning techniques.
• Developed a framework based on clonal group (CG) representation and CG-level features.
• Compared logistic regression, LASSO logistic regression, random forest, and support vector machine models.
• Analyzed antibody repertoire data to identify predictors of CSR.

Main Results:

• The proposed framework can predict CSR events using CG-level features.
• Achieved an unweighted average recall using variable region descriptors and CG diversity measures.
• Successful prediction of CSR was demonstrated both during and before immune challenges.

Conclusions:

• Machine learning models can effectively predict B-cell class switch recombination.
• The CG-based framework offers a novel approach to analyzing antibody repertoire data for immune response studies.
• This predictive capability is valuable for understanding immunological challenges and antibody production.