Sequencing, analyzing, and modeling small samples from large T cell repertoires

Characterizing T cell repertoires is challenging, because repertoires are much larger (i.e. more diverse) than the samples that are sequenced. Additionally, TCRs may differ from each other by as little as a single nucleotide, making it difficult to distinguish erroneous sequences from genuine TCRs. We developed a computational pipeline, called Recover TCR (RTCR), specialized in the complete and accurate retrieval of highly diverse TCR repertoires from high throughput sequencing (HTS) data (Chapter 2). RTCR uses a statistical model to correct PCR- and sequencing errors, and it estimates appropriate parameters for the error correction from the data (i.e., a “data-driven” approach). Next, we apply this pipeline to longitudinal HTS data from blood samples of a healthy volunteer, investigating repertoire dynamics when there is no apparent infection or other trigger for an immune response (Chapter 3). We find that the frequencies of large memory TCRB clonotypes fluctuate over time, presumably due to mounted immune responses. Next, to understand how a diverse naive repertoire is maintained, we develop mathematical models describing the clone-size distribution of the naive repertoire. Surprisingly, we find that naive clone-sizes are to a large extent determined by VDJ recombination probabilities (Chapter 4). Finally, using HTS data processed with the RTCR pipeline, we describe the expressed TRB V, D, J, and C genes in the Ferret, allowing for more detailed future research into the adaptive T cell response of this animal model (Chapter 5).