Abstract
The human naive T-cell receptor (TCR) repertoire is extremely diverse and accurately estimating its distribution is challenging. We address this challenge by combining a quantitative sequencing protocol of TCRA and TCRB sequences with computational modelling. We observed the vast majority of TCR chains only once in our samples, confirming the enormous diversity of the naive repertoire. However, a substantial number of sequences were observed multiple times within samples, and we demonstrated that this is due to expression by many cells in the naive pool. We reason that α and β chains are frequently observed due to a combination of selective processes and summation over multiple clones expressing these chains. We test the contribution of both mechanisms by predicting samples from phenomenological and mechanistically modelled repertoire distributions. By comparing these with sequencing data, we show that frequently observed chains are likely to be derived from multiple clones. Still, a neutral model of T-cell homeostasis cannot account for the observed distributions. We conclude that the data are only compatible with distributions of many small clones in combination with a sufficient number of very large naive T-cell clones, the latter most likely as a result of peripheral selection.
