vaccines and immunoinformatics approaches

The paper develops AIRRTM, a generative model that identifies and generates disease-associated adaptive immune receptor sequences using only repertoire-level immune status data.

This study designs a pan-Salmonella multi-epitope vaccine using subtractive proteomics and immunoinformatics, identifying CsgA as a conserved vaccine candidate with strong immunogenic properties.

SPACE2 improves epitope clustering by using structural models, showing higher coverage and diversity in antibodies binding the same epitope compared to sequence-based methods.

RapTCR provides a tool for rapid visualization and analysis of T-cell receptor repertoires, capturing both global and local sequence similarities.

This paper introduces diffRBM, a transfer-learning model using Restricted Boltzmann Machines to predict antigen immunogenicity and T-cell receptor specificity, achieving performance comparable to existing methods.

This study uses deep unsupervised learning methods to identify and characterize TCR specificity to Sars-Cov-2, achieving promising results in understanding TCR patterns.

The paper fine-tunes protein language models on peptide-MHC sequences to predict MHC-peptide binding, outperforming the existing NetMHCpan4.1 method.

The paper uses template-based modeling to generate TCR:antigen:MHC structures and identifies amino acid contact features that influence antigen recognition.

The paper designs a multi-epitope vaccine against Brucella T4SS by identifying and combining T and B cell epitopes from VirB8 and VirB10 using bioinformatics and molecular simulations.

This study introduces TCRBagger, a weakly-supervised framework for peptide-TCR binding prediction to improve neo-epitope identification in tumor immunotherapy.

The paper trains a transformer model to predict MHCII peptide presentation in humans and mice, showing improved performance with multi-organism data and enabling in silico prediction in a type 1 diabetes model.

The paper introduces BigMHC, a deep learning model that predicts MHC-I epitopes and immunogenic neoepitopes with high accuracy, outperforming existing methods.

This paper reviews proteasomal cleavage prediction methods, benchmarks deep learning architectures, and finds that dataset quality limits performance more than model choice.

DeepTAP predicts TAP-binding peptides using an RNN, outperforming existing methods on benchmark datasets.

This study uses interpretable deep learning to analyze TCR-epitope interactions and identify molecular binding patterns that influence TCR affinity.

This study designs a multi-epitope vaccine against HTLV using molecular dynamics and docking analyses, showing strong TLR binding and stability.

The paper introduces HLAAssoc-1.0 to infer HLA-DRB3/4/5 alleles from HLA-DRB1 data, improving allele and motif deconvolutions in immunopetidomics and training of binding prediction models.

The paper benchmarks 13 HLA genotyping tools on three datasets to identify the most accurate methods and provides consensus predictions for TCGA samples.

STAPLER is a transformer model that predicts TCR-pMHC specificity by learning from TCR and peptide sequences, outperforming existing methods in antigen recognition prediction.

The paper introduces a deep learning model called CcBHLA for predicting HLA class I peptide binding, showing improved performance over existing methods.

The paper improves HLA-DQ antigen presentation prediction using machine learning and shows trans-only variants have limited impact on the immunopeptidome.

The paper compares TCR-pMHC prediction models and finds they lack generalization and robustness due to data dependency.

The paper introduces context-aware amino acid embeddings for TCR sequences, showing improved performance in binding affinity prediction and clustering compared to existing methods.

The paper predicts MHC anchor positions for 328 HLA alleles and shows that incorporating these improves neoantigen identification by correcting misclassified candidates in tumor samples.

This paper develops methods to predict IL-5 inducing peptides using alignment-based and alignment-free approaches, achieving high performance with a hybrid model.

GraphBepi predicts B-cell epitopes using AlphaFold2 structures and a pretrained language model, outperforming existing methods in AUC and AUPR.

TCR2vec is a deep learning framework that learns representations of full T-cell receptors, outperforming methods that focus only on CDR3 regions in predicting binding specificity and clustering.

The paper introduces epiNB, an interpretable machine learning model for predicting patient-specific HLA-I antigen presentation using Mass Spectrometry data, achieving state-of-the-art accuracy with fewer parameters and no hyperparameter tuning.

epitope1D predicts linear B-cell epitopes using explainable machine learning with graph-based descriptors and organism ontology, achieving high accuracy on cross-validation and blind tests.

The paper fine-tunes AlphaFold to predict peptide binding specificity for MHC and other domains, achieving performance comparable to sequence-based methods.