TCRconv: predicting recognition between T cell receptors and epitopes using contextualized motifs

Predictability of antigen binding based on short motifs in the antibody CDRH3

Adaptive immune receptors, such as antibodies and T-cell receptors, recognize foreign threats with exquisite specificity. A major challenge in adaptive immunology is discovering the rules governing immune receptor-antigen binding in order to predict the antigen binding status of previously unseen immune receptors. Many studies assume that the antigen binding status of an immune receptor may be determined by the presence of a short motif in the complementarity determining region 3 (CDR3), disregarding other amino acids. To test this assumption, we present a method to discover short motifs which show high precision in predicting antigen binding and generalize well to unseen simulated and experimental data. Our analysis of a mutagenesis-based antibody dataset reveals 11 336 position-specific, mostly gapped motifs of 3-5 amino acids that retain high precision on independently generated experimental data. Using a subset of only 178 motifs, a simple classifier was made that on the independently generated dataset outperformed a deep learning model proposed specifically for such datasets. In conclusion, our findings support the notion that for some antibodies, antigen binding may be largely determined by a short CDR3 motif. As more experimental data emerge, our methodology could serve as a foundation for in-depth investigations into antigen binding signals.

TCR-H: explainable machine learning prediction of T-cell receptor epitope binding on unseen datasets

Artificial-intelligence and machine-learning (AI/ML) approaches to predicting T-cell receptor (TCR)-epitope specificity achieve high performance metrics on test datasets which include sequences that are also part of the training set but fail to generalize to test sets consisting of epitopes and TCRs that are absent from the training set, i.e., are 'unseen' during training of the ML model. We present TCR-H, a supervised classification Support Vector Machines model using physicochemical features trained on the largest dataset available to date using only experimentally validated non-binders as negative datapoints. TCR-H exhibits an area under the curve of the receiver-operator characteristic (AUC of ROC) of 0.87 for epitope 'hard splitting' (i.e., on test sets with all epitopes unseen during ML training), 0.92 for TCR hard splitting and 0.89 for 'strict splitting' in which neither the epitopes nor the TCRs in the test set are seen in the training data. Furthermore, we employ the SHAP (Shapley additive explanations) eXplainable AI (XAI) method for post hoc interrogation to interpret the models trained with different hard splits, shedding light on the key physiochemical features driving model predictions. TCR-H thus represents a significant step towards general applicability and explainability of epitope:TCR specificity prediction.

Development and Clinical Applications of Therapeutic Cancer Vaccines with Individualized and Shared Neoantigens

Neoantigens, presented as peptides on the surfaces of cancer cells, have recently been proposed as optimal targets for immunotherapy in clinical practice. The promising outcomes of neoantigen-based cancer vaccines have inspired enthusiasm for their broader clinical applications. However, the individualized tumor-specific antigens (TSA) entail considerable costs and time due to the variable immunogenicity and response rates of these neoantigens-based vaccines, influenced by factors such as neoantigen response, vaccine types, and combination therapy. Given the crucial role of neoantigen efficacy, a number of bioinformatics algorithms and pipelines have been developed to improve the accuracy rate of prediction through considering a series of factors involving in HLA-peptide-TCR complex formation, including peptide presentation, HLA-peptide affinity, and TCR recognition. On the other hand, shared neoantigens, originating from driver mutations at hot mutation spots (e.g., KRASG12D), offer a promising and ideal target for the development of therapeutic cancer vaccines. A series of clinical practices have established the efficacy of these vaccines in patients with distinct HLA haplotypes. Moreover, increasing evidence demonstrated that a combination of tumor associated antigens (TAAs) and neoantigens can also improve the prognosis, thus expand the repertoire of shared neoantigens for cancer vaccines. In this review, we provide an overview of the complex process involved in identifying personalized neoantigens, their clinical applications, advances in vaccine technology, and explore the therapeutic potential of shared neoantigen strategies.

Deep learning predictions of TCR-epitope interactions reveal epitope-specific chains in dual alpha T cells

T cells have the ability to eliminate infected and cancer cells and play an essential role in cancer immunotherapy. T cell activation is elicited by the binding of the T cell receptor (TCR) to epitopes displayed on MHC molecules, and the TCR specificity is determined by the sequence of its α and β chains. Here, we collect and curate a dataset of 17,715 αβTCRs interacting with dozens of class I and class II epitopes. We use this curated data to develop MixTCRpred, an epitope-specific TCR-epitope interaction predictor. MixTCRpred accurately predicts TCRs recognizing several viral and cancer epitopes. MixTCRpred further provides a useful quality control tool for multiplexed single-cell TCR sequencing assays of epitope-specific T cells and pinpoints a substantial fraction of putative contaminants in public databases. Analysis of epitope-specific dual α T cells demonstrates that MixTCRpred can identify α chains mediating epitope recognition. Applying MixTCRpred to TCR repertoires from COVID-19 patients reveals enrichment of clonotypes predicted to bind an immunodominant SARS-CoV-2 epitope. Overall, MixTCRpred provides a robust tool to predict TCRs interacting with specific epitopes and interpret TCR-sequencing data from both bulk and epitope-specific T cells.

Prediction of tumor-reactive T cell receptors from scRNA-seq data for personalized T cell therapy

The identification of patient-derived, tumor-reactive T cell receptors (TCRs) as a basis for personalized transgenic T cell therapies remains a time- and cost-intensive endeavor. Current approaches to identify tumor-reactive TCRs analyze tumor mutations to predict T cell activating (neo)antigens and use these to either enrich tumor infiltrating lymphocyte (TIL) cultures or validate individual TCRs for transgenic autologous therapies. Here we combined high-throughput TCR cloning and reactivity validation to train predicTCR, a machine learning classifier that identifies individual tumor-reactive TILs in an antigen-agnostic manner based on single-TIL RNA sequencing. PredicTCR identifies tumor-reactive TCRs in TILs from diverse cancers better than previous gene set enrichment-based approaches, increasing specificity and sensitivity (geometric mean) from 0.38 to 0.74. By predicting tumor-reactive TCRs in a matter of days, TCR clonotypes can be prioritized to accelerate the manufacture of personalized T cell therapies.

Single-cell analysis of immune recognition in chronic myeloid leukemia patients following tyrosine kinase inhibitor discontinuation

Immunological control of residual leukemia cells is thought to occur in patients with chronic myeloid leukemia (CML) that maintain treatment-free remission (TFR) following tyrosine kinase inhibitor (TKI) discontinuation. To study this, we analyzed 55 single-cell RNA and T cell receptor (TCR) sequenced samples (scRNA+TCRαβ-seq) from patients with CML (n = 13, N = 25), other cancers (n = 28), and healthy (n = 7). The high number and active phenotype of natural killer (NK) cells in CML separated them from healthy and other cancers. Most NK cells in CML belonged to the active CD56dim cluster with high expression of GZMA/B, PRF1, CCL3/4, and IFNG, with interactions with leukemic cells via inhibitory LGALS9-TIM3 and PVR-TIGIT interactions. Accordingly, upregulation of LGALS9 was observed in CML target cells and TIM3 in NK cells when co-cultured together. Additionally, we created a classifier to identify TCRs targeting leukemia-associated antigen PR1 and quantified anti-PR1 T cells in 90 CML and 786 healthy TCRβ-sequenced samples. Anti-PR1 T cells were more prevalent in CML, enriched in bone marrow samples, and enriched in the mature, cytotoxic CD8 + TEMRA cluster, especially in a patient maintaining TFR. Our results highlight the role of NK cells and anti-PR1 T cells in anti-leukemic immune responses in CML.

Systems immunology spanning tumors, lymph nodes, and periphery

The immune system defines a complex network of tissues and cell types that orchestrate responses across the body in a dynamic manner. The local and systemic interactions between immune and cancer cells contribute to disease progression. Lymphocytes are activated in lymph nodes, traffic through the periphery, and impact cancer progression through their interactions with tumor cells. As a result, therapeutic response and resistance are mediated across tissues, and a comprehensive understanding of lymphocyte dynamics requires a systems-level approach. In this review, we highlight experimental and computational methods that can leverage the study of leukocyte trafficking through an immunomics lens and reveal how adaptive immunity shapes cancer.

EPIC-TRACE: predicting TCR binding to unseen epitopes using attention and contextualized embeddings

MOTIVATION

T cells play an essential role in adaptive immune system to fight pathogens and cancer but may also give rise to autoimmune diseases. The recognition of a peptide-MHC (pMHC) complex by a T cell receptor (TCR) is required to elicit an immune response. Many machine learning models have been developed to predict the binding, but generalizing predictions to pMHCs outside the training data remains challenging.

RESULTS

We have developed a new machine learning model that utilizes information about the TCR from both α and β chains, epitope sequence, and MHC. Our method uses ProtBERT embeddings for the amino acid sequences of both chains and the epitope, as well as convolution and multi-head attention architectures. We show the importance of each input feature as well as the benefit of including epitopes with only a few TCRs to the training data. We evaluate our model on existing databases and show that it compares favorably against other state-of-the-art models.

AVAILABILITY AND IMPLEMENTATION

https://github.com/DaniTheOrange/EPIC-TRACE.

Quantitative approaches for decoding the specificity of the human T cell repertoire

T cell receptor (TCR)-peptide-major histocompatibility complex (pMHC) interactions play a vital role in initiating immune responses against pathogens, and the specificity of TCRpMHC interactions is crucial for developing optimized therapeutic strategies. The advent of high-throughput immunological and structural evaluation of TCR and pMHC has provided an abundance of data for computational approaches that aim to predict favorable TCR-pMHC interactions. Current models are constructed using information on protein sequence, structures, or a combination of both, and utilize a variety of statistical learning-based approaches for identifying the rules governing specificity. This review examines the current theoretical, computational, and deep learning approaches for identifying TCR-pMHC recognition pairs, placing emphasis on each method's mathematical approach, predictive performance, and limitations.

CrossDome: an interactive R package to predict cross-reactivity risk using immunopeptidomics databases

T-cell-based immunotherapies hold tremendous potential in the fight against cancer, thanks to their capacity to specifically targeting diseased cells. Nevertheless, this potential has been tempered with safety concerns regarding the possible recognition of unknown off-targets displayed by healthy cells. In a notorious example, engineered T-cells specific to MAGEA3 (EVDPIGHLY) also recognized a TITIN-derived peptide (ESDPIVAQY) expressed by cardiac cells, inducing lethal damage in melanoma patients. Such off-target toxicity has been related to T-cell cross-reactivity induced by molecular mimicry. In this context, there is growing interest in developing the means to avoid off-target toxicity, and to provide safer immunotherapy products. To this end, we present CrossDome, a multi-omics suite to predict the off-target toxicity risk of T-cell-based immunotherapies. Our suite provides two alternative protocols, i) a peptide-centered prediction, or ii) a TCR-centered prediction. As proof-of-principle, we evaluate our approach using 16 well-known cross-reactivity cases involving cancer-associated antigens. With CrossDome, the TITIN-derived peptide was predicted at the 99+ percentile rank among 36,000 scored candidates (p-value < 0.001). In addition, off-targets for all the 16 known cases were predicted within the top ranges of relatedness score on a Monte Carlo simulation with over 5 million putative peptide pairs, allowing us to determine a cut-off p-value for off-target toxicity risk. We also implemented a penalty system based on TCR hotspots, named contact map (CM). This TCR-centered approach improved upon the peptide-centered prediction on the MAGEA3-TITIN screening (e.g., from 27th to 6th, out of 36,000 ranked peptides). Next, we used an extended dataset of experimentally-determined cross-reactive peptides to evaluate alternative CrossDome protocols. The level of enrichment of validated cases among top 50 best-scored peptides was 63% for the peptide-centered protocol, and up to 82% for the TCR-centered protocol. Finally, we performed functional characterization of top ranking candidates, by integrating expression data, HLA binding, and immunogenicity predictions. CrossDome was designed as an R package for easy integration with antigen discovery pipelines, and an interactive web interface for users without coding experience. CrossDome is under active development, and it is available at https://github.com/AntunesLab/crossdome.