DeepPROTECTNeo: A Deep learning-based Personalized and RV-guided Optimization tool for TCR Epitope interaction using Context-aware Transformers

The development of personalized cancer vaccines relies heavily on accurately identifying neoepitopes capable of eliciting strong immune responses. Traditional methods primarily focus on predicting interactions between major histocompatibility complex (MHC) molecules and peptides, often neglecting the crucial step of T-cell receptor (TCR) binding in the context of anchor motif and hydrophobicity, which are critical variables in the elicitation of robust tumor-specific responses. T cells perpetually monitor lymphoid and peripheral organs for antigens, such as peptides or lipids, that are presented by peptide-MHC (pMHC) molecules in other cells. This oversight limits the efficacy of these vaccines in triggering a robust adaptive immune response. We introduce a stand-alone web-server DeepPROTECTNeo ( https://cosmos.iitkgp.ac.in/DeepPROTECTNeo/ ) that processes Whole Exome Sequencing (WES) or Whole Genome Sequencing (WGS) data and mitigate the process of extracting patient-specific neoantigen candidate leveraging genomic and variational TCR information directly, a pivotal component in the development of adaptive immunity and personalized cancer vaccines. Along with traditional MHC binding and B-cell receptor (BCR) recognition, we introduce a context-aware transformer block leveraging cross-attention mechanisms to enhance the analysis of TCR-epitope interactions, thereby advancing neoepitope discovery with unparalleled precision and facilitating reverse vaccinology (RV)-guided multi-epitope vaccine design. With extensive hyperparameter tuning, our model shows exceptional generalizability in predicting TCR-peptide binding on independent test data with AUC 0.7234 and AUPRC 0.7659, substantially outperforming state-of-the-art methods. Furthermore, our pipeline successfully identified 372 neoepitopes out of 532 unique neoepitopes presented in the TESLA benchmark dataset, with 24 true neoantigens, demonstrating its high efficacy and precision in neoepitope discovery.