Inferring mechanisms of V(D)J recombination using statistical inference on high-throughput immune repertoire data

Abstract

To appropriately defend against a wide array of pathogens, jawed vertebrates somatically generate highly diverse repertoires of B cell and T cell receptors through a random process called V(D)J recombination. Receptor diversity arises during recombination from the combinatorial assembly of V(D)J genes and the junctional deletion and insertion of nucleotides. While molecular experiments have established our understanding of V(D)J recombination in vitro, the processes underlying receptor generation in vivo, particularly in humans with intact recombination machinery, remain poorly characterized. This dissertation uses statistical inference on large immune receptor repertoire sequencing datasets to investigate the molecular mechanisms of V(D)J recombination in humans, with a focus on individual variability, nucleotide trimming, and the role of sequence microhomology. First, I identify genetic loci associated with modifying V(D)J recombination probabilities using genome-wide association inference and reveal individual differences in receptor generation. Next, I develop a probabilistic model of nucleotide trimming to infer how sequence-level features influence this process. Finally, I demonstrate that germline-encoded microhomology biases both trimming and ligation outcomes, providing mechanistic insights into its role in recombination. Together, these findings advance our understanding of how immune receptordiversity is generated and establish a foundation for future research on individual variability in immune responses.