Bayesian Inference of Allelic Inclusion Rates in the Human T Cell Receptor Repertoire.

Abstract

A small population of αβ Tcells is characterized by the expression of more than one unique Tcell receptor (TCR); this outcome is the result of "allelic inclusion," that is, inclusion of both α- or β-chain alleles during V(D)J recombination. Limitations in single-cell sequencing technology, however, have precluded comprehensive enumeration of these dual receptor Tcells. Here, we develop and experimentally validate a fully Bayesian inference model capable of reliably estimating the true rate of α and β TCR allelic inclusion across two different emulsion-barcoding single-cell sequencing platforms. We provide a database composed of over 51,000 previously unpublished allelic inclusion TCR sequence sets drawn from eight healthy individuals and show that allelic inclusion contributes a distinct and functionally important set of sequences to the human TCR repertoire. This database and a Python implementation of our statistical inference model are freely available at our Github repository (https://github.com/JasonACarter/Allelic_inclusion).