Quantitative annotations of T-Cell repertoire specificity

Identifying T-cell clubs by embracing the local harmony between TCR and gene expressions

T cell receptors (TCR) and gene expression provide two complementary and essential aspects in T cell understanding, yet their diversity presents challenges in integrative analysis. We introduce TCRclub, a novel method integrating single-cell RNA sequencing data and single-cell TCR sequencing data using local harmony to identify functionally similar T cell groups, termed 'clubs'. We applied TCRclub to 298,106 T cells across seven datasets encompassing various diseases. First, TCRclub outperforms the state-of-the-art methods in clustering T cells on a dataset with over 400 verified peptide-major histocompatibility complex categories. Second, TCRclub reveals a transition from activated to exhausted T cells in cholangiocarcinoma patients. Third, TCRclub discovered the pathways that could intervene in response to anti-PD-1 therapy for patients with basal cell carcinoma by analyzing the pre-treatment and post-treatment samples. Furthermore, TCRclub unveiled different T-cell responses and gene patterns at different severity levels in patients with COVID-19. Hence, TCRclub aids in developing more effective immunotherapeutic strategies for cancer and infectious diseases.

A comprehensive benchmarking for evaluating TCR embeddings in modeling TCR-epitope interactions

The complexity of T cell receptor (TCR) sequences, particularly within the complementarity-determining region 3 (CDR3), requires efficient embedding methods for applying machine learning to immunology. While various TCR CDR3 embedding strategies have been proposed, the absence of their systematic evaluations created perplexity in the community. Here, we extracted CDR3 embedding models from 19 existing methods and benchmarked these models with four curated datasets by accessing their impact on the performance of TCR downstream tasks, including TCR-epitope binding affinity prediction, epitope-specific TCR identification, TCR clustering, and visualization analysis. We assessed these models utilizing eight downstream classifiers and five downstream clustering methods, with the performance measured by a diverse range of metrics for precision, robustness, and usability. Overall, handcrafted embeddings outperformed data-driven ones in modeling TCR-epitope interactions. To further refine our comparative findings, we developed an all-in-one TCR CDR3 embedding package comprising all evaluated embedding models. This package will assist users in easily selecting suitable embedding models for their data.