Attentive Variational Information Bottleneck for TCR–peptide interaction prediction

Machart, Pierre; Mösch, Anja; Li, Kai; Castorina, Leonardo V.; Pfeifer, Nico; Min, Martin Renqiang

doi:10.1093/bioinformatics/btac820

Cited by 10 publications

(12 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For our experiments, two state-of-the-art deep learning models for TCR-peptide interaction prediction: NetTCR-2.0 Jurtz et al (2018) and AVIB Grazioli et al (2022a). We evaluated how their performance is affected by training and testing on different data splits.…”

Section: Resultsmentioning

confidence: 99%

“…To generate non-binding (i.e., negative) samples, we randomly shuffled the available (peptide, CDR3- β ) pairs. This process is commonly employed in the literature (Jurtz et al, 2018; Grazioli et al, 2022a) and leverages the hypothesis that random pairs will most likely not bind. To create a balanced dataset, we randomly generated 36,641 samples of non-binding combinations of CDR3- β and peptide sequences, increasing the total number of data points to 65,946.…”

Section: Methodsmentioning

confidence: 99%

“…The development of computational models that predict whether TCRs and pMHCs interact would allow for a faster in silico screening of sequences and consequent design of TCR sequences that bind specific target pMHCs. Recent advancements in machine learning (ML) lead to the development of TCR binding predictors (Springer et al, 2021; Montemurro et al, 2021; Weber et al, 2021; Grazioli et al, 2022a). The data input of these models consists of tuples of short amino acid sequences such as (peptide, CDR3- β ) or (peptide, CDR3- β , CDR3- α , MHC).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Assessing the Generalization Capabilities of TCR Binding Predictors via Peptide Distance Analysis

Castorina,

Grazioli,

Machart

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Accurate TCR-peptide/pMHC binding prediction is crucial to understand immune responses and develop therapeutic interventions. In this study, we propose a novel approach to assess the generalization capabilities of machine learning models in predicting TCR-peptide binding for unseen viral peptides. Our method involves measuring the distance between training and test peptides and leveraging this metric to assess model generalization capabilities. In this study, we employ sequence-based and structure-based distances. Through extensive experimentation and analysis, we demonstrate that the distance-based assessment provides valuable insights into the generalizability of the model and its ability to predict TCR binding to previously unobserved viral peptides. The findings from this study shed light on the challenges and opportunities in TCR-peptide/pMHC binding prediction and contribute to the advancement of computational immunology and peptide-based therapeutics.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Assessing the Generalization Capabilities of TCR Binding Predictors via Peptide Distance Analysis

Castorina,

Grazioli,

Machart

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…A more complex modification would be the simulation of the tumor microenvironment, especially T cell response. Diversity in T cell repertoire could be used as a proxy ( Azizi et al 2018 , Han et al 2020 ) or even prediction of neoantigen-specific T cell receptors ( Montemurro et al 2022 , Grazioli et al 2023 ), although this is currently not a feasible approach due to the limited reliability of current TCR binding prediction algorithms ( Grazioli et al 2022 ).…”

Section: Discussionmentioning

confidence: 99%

NeoAgDT: optimization of personal neoantigen vaccine composition by digital twin simulation of a cancer cell population

Mösch,

Grazioli,

Machart

et al. 2024

Bioinformatics

Self Cite

View full text Add to dashboard Cite

Motivation Neoantigen vaccines make use of tumor-specific mutations to enable the patient’s immune system to recognize and eliminate cancer. Selecting vaccine elements, however, is a complex task which needs to take into account not only the underlying antigen presentation pathway but also tumor heterogeneity. Results Here, we present NeoAgDT, a two-step approach consisting of: (1) simulating individual cancer cells to create a digital twin of the patient’s tumor cell population and (2) optimizing the vaccine composition by integer linear programming based on this digital twin. NeoAgDT shows improved selection of experimentally-validated neoantigens over ranking-based approaches in a study of seven patients. Availability The NeoAgDT code is published on Github: https://github.com/nec-research/neoagdt Supplementary information Supplementary data are available at Bioinformatics online.

show abstract

“…There are simple models such as Bayesian non-parametric model [1], Random Forest (TCRex [2], epiTCR [3]), and clustering-based models (TCRdist [4,5]). More complex models [5,6,7,8] are also proposed for the classification task. Many deep learning models (NetTCR [9], DeepTCR [7], ImRex [8], tcrpred [10]) rely on convolutional neural networks to learn the TCR and peptide patterns in each interaction.…”

Section: Introductionmentioning

confidence: 99%

epiTCR-KDA: Knowledge Distillation model on Dihedral Angles for TCR-peptide prediction

Pham,

Su,

Nguyen

et al. 2024

Preprint

View full text Add to dashboard Cite

Motivation: Antigen recognition by T-cell receptors (TCRs) triggers cascades of immune responses. Successful predictions of the TCR and antigen (as peptide) bindings therefore signify the advancements in immunotherapy. However, most of current TCR-peptide interaction predictors fail to predict unseen data. This limitation may be derived from the conventional usage of TCR and/or peptide sequences as input, which may not adequately reflect their structural characteristics. Therefore, incorporating the TCR and peptide structural information into the prediction model to improve the generalizability is necessary. Results: We presented epiTCR-KDA as a new predictor of TCR-peptide binding that utilises structural information, specifically the dihedral angles between the residues of both the peptide and the TCR. This structural descriptor was integrated into a model constructed using knowledge distillation to enhance its generalizability. The epiTCR-KDA demonstrated competitive prediction performance, with an AUC of 0.99 for seen data and AUC of 0.86 for unseen data. Across multiple public datasets, epiTCR-KDA consistently outperformed other predictors, such as epiTCR, NetTCR, BERTrand, TEIM-Seq, TEINet, and ImRex, maintaining a median AUC of 0.9 (ranging from 0.82 to 0.91). Further analysis of epiTCR-KDA performance indicated that the cosine similarity of the dihedral angle vectors between the unseen testing data and training data is crucial for its stable performance. In conclusion, our epiTCR-KDA model, with its capacity to predict for unseen data, has brought us one step closer toward the development of a highly effective pipeline for affordable antigen-based immunotherapy. Availability and implementation: epiTCR-KDA is available on GitHub (https://github.com/ddiem-ri-4D/epiTCR-KDA) .

show abstract

Attentive Variational Information Bottleneck for TCR–peptide interaction prediction

Cited by 10 publications

References 49 publications

Assessing the Generalization Capabilities of TCR Binding Predictors via Peptide Distance Analysis

Assessing the Generalization Capabilities of TCR Binding Predictors via Peptide Distance Analysis

NeoAgDT: optimization of personal neoantigen vaccine composition by digital twin simulation of a cancer cell population

epiTCR-KDA: Knowledge Distillation model on Dihedral Angles for TCR-peptide prediction

Contact Info

Product

Resources

About