Inference of SARS-CoV-2 exposure biomarkers using large-scale T-cell repertoire profiling

Area of Science:

• Immunology
• Infectious Disease Epidemiology
• Bioinformatics

Background:

• The COVID-19 pandemic highlights the need for methods to track infectious disease spread using population immunity signatures.
• Adaptive immune receptor repertoire sequencing (AIRR-seq) identifies T cell receptor (TCR) biomarkers for pathogen specificity and immune memory.
• AIRR-seq detects infection imprints and aids in studying individual SARS-CoV-2 responses.

Purpose of the Study:

• To develop and apply a machine learning approach for inferring SARS-CoV-2 exposure from TCR repertoire features.
• To create a robust classifier for COVID-19 exposure assessment using AIRR-seq data.

Main Methods:

• Applied machine learning to two large AIRR-seq datasets (>1,200 repertoires) from healthy and COVID-19-convalescent donors.
• Utilized a novel batch effect correction method for combining data from different batches and protocols.
• Ensured high-quality data through standardization, human leukocyte antigen (HLA) typing, and analysis of both TCR α- and β-chain sequences.

Main Results:

• Identified specific TCR repertoire features induced by SARS-CoV-2 exposure.
• Developed a robust and highly accurate classifier for predicting COVID-19 exposure.
• Demonstrated the utility of machine learning in analyzing large-scale AIRR-seq datasets.

Conclusions:

• The developed classifier is applicable to individual TCR repertoires generated via various protocols.
• This approach enables AIRR-seq-based immune status assessment in large donor cohorts.
• Facilitates broader application of AIRR-seq for infectious disease monitoring and immunological studies.