Machine learning analysis of the T cell receptor repertoire identifies sequence features of self-reactivity
Johannes Textor, Franka Buytenhuijs, Dakota Rogers, Ève Mallet Gauthier, Shabaz Sultan, Inge M. N. Wortel, Kathrin Kalies, Anke Fähnrich, René Pagel, Heather J. Melichar, Jürgen Westermann, Judith N. Mandl
Abstract
The T cell receptor (TCR) determines specificity and affinity for both foreign and self-peptides presented by the major histocompatibility complex (MHC). Although the strength of TCR interactions with self-pMHC impacts T cell function, it has been challenging to identify TCR sequence features that predict T cell fate. To discern patterns distinguishing TCRs from naive CD4 + T cells with low versus high self-reactivity, we used data from 42 mice to train a machine learning (ML) algorithm that identifies population-level differences between TCR β sequence sets. This approach revealed that weakly self-reactive T cell populations were enriched for longer CDR3 β regions and acidic amino acids. We tested our ML predictions of self-reactivity using retrogenic mice with fixed TCR β sequences. Extrapolating our analyses to independent datasets, we predicted high self-reactivity for regulatory T cells and slightly reduced self-reactivity for T cells responding to chronic infections. Our analyses suggest a potential trade-off between TCR repertoire diversity and self-reactivity. A record of this paper's transparent peer review process is included in the supplemental information.