Novel Ensemble Feature Selection Approach and Application in Repertoire Sequencing Data.

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• The adaptive immune system, comprising T and B cells, is crucial for immune responses.
• Adaptive immunity generation relies on V(D)J gene recombination, making VJ gene usage a potential biomarker.
• Analyzing the adaptive immune system is complex due to repertoire heterogeneity.

Purpose of the Study:

• To develop a novel ensemble feature selection approach and statistical learning algorithm to analyze T and B cell VJ gene usage.
• To address the heterogeneity challenge in adaptive immune repertoire analysis.
• To identify VJ gene signatures associated with COVID-19 recovery and immunotherapy response in lung cancer.

Main Methods:

• Proposed a novel ensemble feature selection approach and a customized statistical learning algorithm.
• Focused on VJ gene usage analysis.
• Applied the approach to T cell receptor sequences from recovered COVID-19 patients, healthy donors, and lung cancer patients undergoing immunotherapy.

Main Results:

• Identified distinct VJ gene usage patterns in recovered COVID-19 patients compared to healthy donors.
• Discovered VJ genes associated with clinical response in lung cancer patients receiving immunotherapy.
• Simulation studies demonstrated superior efficiency, accuracy, stability, and sensitivity of the ensemble approach over existing methods.

Conclusions:

• The proposed ensemble feature selection approach is effective for classifying immune subtypes and understanding adaptive immune responses.
• This method aids in identifying disease- or treatment-associated immune signatures.
• The findings can contribute to improved therapeutic strategies by better understanding immune responses.