Clinical Fellow, Post-Doctoral Research Fellow Yale School of Medicine New Haven, Connecticut, United States
Disclosure(s):
Oyunbileg Magvanjav, MD, PhD: No financial relationships to disclose
Introduction/Rationale: High-throughput bulk and single-cell B cell receptor (BCR) sequencing each provide valuable insights into adaptive immunity. While bulk sequencing achieves greater depth at lower cost, it may overrepresent cell populations with elevated mRNA counts, such as plasma cells. This bias has the potential to skew BCR repertoire features like clonal abundance, somatic hypermutation (SHM), isotype frequency, and V-gene usage.
Methods: We systematically compared BCR repertoire characteristics between bulk and single-cell sequencing data generated from the same samples across four publicly available datasets (influenza vaccination, myasthenia gravis, pediatric tonsillitis, healthy adults), along with newly generated BCR-seq data from this study. We analyzed clonal abundance, IGHV SHM frequency, V-gene usage, and isotype distributions. We also used simulated repertoires with varied mRNA abundances to evaluate the impact of transcriptional differences on biases in repertoire feature estimation.
Results: Bulk sequencing consistently showed higher clonal abundance, particularly in clones containing plasma cells, as identified via single-cell RNA-seq annotations. Simulations confirmed that higher mRNA levels are sufficient to skew clonal abundance estimates in bulk sequencing. Collapsing bulk data to unique sequences attenuated, but did not eliminate, this bias, potentially due to technical factors such as uncorrected sequencing errors. Additional repertoire features showed variable agreement between sequencing methods: IGHV SHM frequency, V-gene usage, and isotype distributions differed in some samples, with differences often associated with sample-specific repertoire composition and cellular heterogeneity.
Conclusion: Our findings show that bulk BCR sequencing introduces systematic biases related to cell-type-specific mRNA abundance. These biases may influence inferred repertoire features and underscore the importance of considering sequencing methodology when interpreting BCR repertoires.
