On the viability of unsupervised T-cell receptor sequence clustering for epitope preference

Abstract

The T-cell receptor is responsible for recognizing potentially harmful epitopes presented on cell surfaces. The binding rules that govern this recognition between receptor and epitope is currently an unsolved problem, yet one of great interest. Several methods have been proposed recently to perform supervised classification of T-cell receptor sequences, but this requires known examples of T-cell sequences for a given epitope. Here we study the viability of various methods to perform unsupervised clustering of distinct T-cell receptor sequences and how these clusters relate to their target epitope. The goal is to provide an overview of the performance of various distance metrics on two large independent T-cell receptor sequence data sets. Our results confirm the presence of structural distinct T-cell groups that target identical epitopes. In addition, we put forward several recommendations to perform T-cell receptor sequence clustering.

Current challenges for epitope-agnostic TCR interaction prediction and a new perspective derived from image classification

“…In essence, these models need to be able to understand the processes that govern the affinity between a TCR and epitope on a molecular level. This process is rather complex, because while sequence similarity seems to be related to the binding process, epitopes can be bound by varying TCRs (10)(11)(12) and TCRs can also display crossreactivity (13). Despite the advent of single cell TCR paired chain sequencing, currently available TCR-epitope binding data mostly consists of single-chain, often beta-chain, data, while in reality both the alpha-and beta-chains are thought to contribute to binding specificity.…”

Preexisting memory CD4 T cells in naïve individuals confer robust immunity upon hepatitis B vaccination

“…These peptide-specific TCRβ sequences can be utilized in a peptide-TCR interaction classifier to identify other TCRβ that are likely to react against the same HBsAg epitopes, as it has been shown that similar TCRβ sequences tend to target the same epitopes 16 Fig. 3b).…”

TCRex: detection of enriched T cell epitope specificity in full T cell receptor sequence repertoires

“…However, we do have some reasons to believe that the epitope-specific prediction models are MHC-agnostic. First, previous studies have shown that TCRs can be shared across individuals with different HLA-types (20) and that epitope-specific TCR patterns seem to transcend the MHC background of the epitopes (11). This might indicate that epitope-specific TCR sequences can recognize a specific epitope even when it is presented by different MHCs.…”

“…To determine the epitope specificity of TCRs, machine learning models have been developed to analyze TCR sequences and predict the probability that they recognize and bind specific epitopes. These methods are based on the principle that similar TCR sequences often target the same epitope (11) and that machine learning techniques can be used to learn the molecular underpinnings that are shared by these epitope-specific TCR sequences (12)(13)(14). While these methods have been shown to be performant on small targeted data sets, their application on full repertoire datasets remains challenging.…”