{ "source": "supplemental material.docx", "generated_from": "supplemental material.docx tables S2-S5", "categories": [ { "key": "sequence", "label": "Sequence similarity and off-target prioritization", "table": "Table S2", "description": "Sequence similarity, database matching, and peptide mimicry tools for cross-reactivity or off-target candidate prioritization." }, { "key": "structure", "label": "Structure-based modeling and candidate refinement", "table": "Table S3", "description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation." }, { "key": "ml_dl", "label": "Machine learning / deep learning specificity models", "table": "Table S4", "description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery." }, { "key": "multimodal", "label": "Multimodal cross-reactivity risk and engineering", "table": "Table S5", "description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation." } ], "tools": [ { "id": "sequence-001", "name": "Expitope", "year": 2015, "method": "Local sequence alignment + tissue-expression profiling", "methodology": "Local sequence alignment + tissue-expression profiling", "input": "Peptide; user-defined mismatch tolerance", "data_source": "In-house dataset", "prediction_task": "Off-target toxicity risk assessment", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[13]", "ref": "[13]", "category": "sequence", "category_label": "Sequence similarity and off-target prioritization", "category_description": "Sequence similarity, database matching, and peptide mimicry tools for cross-reactivity or off-target candidate prioritization.", "supplemental_table": "Table S2", "brief_description": "Off-target toxicity risk assessment Method: Local sequence alignment + tissue-expression profiling. Input: Peptide; user-defined mismatch tolerance.", "link": "", "Citation_Count": "" }, { "id": "sequence-002", "name": "TCRMatch", "year": 2021, "method": "k-mer exact matching", "methodology": "k-mer exact matching", "input": "CDR3β", "data_source": "IEDB", "prediction_task": "TCR specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[14]", "ref": "[14]", "category": "sequence", "category_label": "Sequence similarity and off-target prioritization", "category_description": "Sequence similarity, database matching, and peptide mimicry tools for cross-reactivity or off-target candidate prioritization.", "supplemental_table": "Table S2", "brief_description": "TCR specificity prediction Method: k-mer exact matching. Input: CDR3β.", "link": "", "Citation_Count": "" }, { "id": "sequence-003", "name": "ProtBer", "year": 2023, "method": "Sequence embedding + network-based relational modeling", "methodology": "Sequence embedding + network-based relational modeling", "input": "CDR3α/β", "data_source": "-", "prediction_task": "Immune-response pattern classification", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[15]", "ref": "[15]", "category": "sequence", "category_label": "Sequence similarity and off-target prioritization", "category_description": "Sequence similarity, database matching, and peptide mimicry tools for cross-reactivity or off-target candidate prioritization.", "supplemental_table": "Table S2", "brief_description": "Immune-response pattern classification Method: Sequence embedding + network-based relational modeling. Input: CDR3α/β.", "link": "", "Citation_Count": "" }, { "id": "sequence-004", "name": "MORITS", "year": 2024, "method": "Weighted sequence alignment emphasizing MHC outward-facing residues", "methodology": "Weighted sequence alignment emphasizing MHC outward-facing residues", "input": "Peptide", "data_source": "-", "prediction_task": "Cross-reactivity prediction for heterologous peptides", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[16]", "ref": "[16]", "category": "sequence", "category_label": "Sequence similarity and off-target prioritization", "category_description": "Sequence similarity, database matching, and peptide mimicry tools for cross-reactivity or off-target candidate prioritization.", "supplemental_table": "Table S2", "brief_description": "Cross-reactivity prediction for heterologous peptides Method: Weighted sequence alignment emphasizing MHC outward-facing residues. Input: Peptide.", "link": "", "Citation_Count": "" }, { "id": "structure-005", "name": "TCRmodel", "year": 2018, "method": "Rosetta-based structural modeling", "methodology": "Rosetta-based structural modeling", "input": "TCR Vα/Vβ", "data_source": "PDB; VDJdb", "prediction_task": "TCR structural prediction", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[17]", "ref": "[17]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "TCR structural prediction Method: Rosetta-based structural modeling. Input: TCR Vα/Vβ.", "link": "", "Citation_Count": "" }, { "id": "structure-006", "name": "HLA-Arena", "year": 2020, "method": "Structural modeling and computational interface analysis", "methodology": "Structural modeling and computational interface analysis", "input": "Peptide; HLA allele", "data_source": "PDB; IEDB; in-house data", "prediction_task": "pMHC structural modeling and interface evaluation", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[18]", "ref": "[18]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "pMHC structural modeling and interface evaluation Method: Structural modeling and computational interface analysis. Input: Peptide; HLA allele.", "link": "", "Citation_Count": "" }, { "id": "structure-007", "name": "RACER", "year": 2021, "method": "Supervised machine learning + energy-based optimization", "methodology": "Supervised machine learning + energy-based optimization", "input": "CDR3α/β; peptide; available structural templates", "data_source": "Experimentally derived data", "prediction_task": "TCR–peptide affinity estimation", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[19]", "ref": "[19]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "TCR–peptide affinity estimation Method: Supervised machine learning + energy-based optimization. Input: CDR3α/β; peptide; available structural templates.", "link": "", "Citation_Count": "" }, { "id": "structure-008", "name": "Modeller", "year": 2022, "method": "Template alignment + energy optimization", "methodology": "Template alignment + energy optimization", "input": "CDR3α/β; pMHC; user-specified PDB templates", "data_source": "PDB; user-provided templates", "prediction_task": "TCR–pMHC complex modeling", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[20]", "ref": "[20]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "TCR–pMHC complex modeling Method: Template alignment + energy optimization. Input: CDR3α/β; pMHC; user-specified PDB templates.", "link": "", "Citation_Count": "" }, { "id": "structure-009", "name": "AlphaFold", "year": 2023, "method": "Deep learning–based structure prediction", "methodology": "Deep learning–based structure prediction", "input": "TCR α/β; pMHC", "data_source": "PDB", "prediction_task": "TCR–pMHC structural prediction and binding assessment", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[21]", "ref": "[21]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "TCR–pMHC structural prediction and binding assessment Method: Deep learning–based structure prediction. Input: TCR α/β; pMHC.", "link": "", "Citation_Count": "" }, { "id": "structure-010", "name": "TCRmodel2", "year": 2023, "method": "AlphaFold-optimized model with TCR-specific constraints", "methodology": "AlphaFold-optimized model with TCR-specific constraints", "input": "TCR α/β; pMHC", "data_source": "UniProt; BFD", "prediction_task": "TCR–pMHC structural prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[22]", "ref": "[22]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "TCR–pMHC structural prediction Method: AlphaFold-optimized model with TCR-specific constraints. Input: TCR α/β; pMHC.", "link": "", "Citation_Count": "" }, { "id": "structure-011", "name": "TCRen", "year": 2024, "method": "Structure-based statistical-potential scoring with modeled TCR–pMHC complexes", "methodology": "Structure-based statistical-potential scoring with modeled TCR–pMHC complexes", "input": "TCR Vα/Vβ, peptide, MHC", "data_source": "PDB; modeled TCR–pMHC structures", "prediction_task": "Ranking candidate unseen epitopes / TCR recognition prediction", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[23]", "ref": "[23]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "Ranking candidate unseen epitopes / TCR recognition prediction Method: Structure-based statistical-potential scoring with modeled TCR–pMHC complexes. Input: TCR Vα/Vβ, peptide, MHC.", "link": "", "Citation_Count": "" }, { "id": "structure-012", "name": "MatchTope", "year": 2022, "method": "MEP analysis + hierarchical clustering", "methodology": "MEP analysis + hierarchical clustering", "input": "pMHC structural coordinates", "data_source": "Viral immunology datasets", "prediction_task": "pMHC cross-reactivity prediction", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[24]", "ref": "[24]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "pMHC cross-reactivity prediction Method: MEP analysis + hierarchical clustering. Input: pMHC structural coordinates.", "link": "", "Citation_Count": "" }, { "id": "structure-013", "name": "TCRpcDist", "year": 2024, "method": "3D structural similarity + physicochemical interface profiling", "methodology": "3D structural similarity + physicochemical interface profiling", "input": "CDR1/2/3 (α/β)", "data_source": "PDB; VDJdb", "prediction_task": "Peptide specificity inference through structural analogs", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[25]", "ref": "[25]", "category": "structure", "category_label": "Structure-based modeling and candidate refinement", "category_description": "Structural modeling, interface analysis, and candidate refinement tools for TCR-pMHC or pMHC interpretation.", "supplemental_table": "Table S3", "brief_description": "Peptide specificity inference through structural analogs Method: 3D structural similarity + physicochemical interface profiling. Input: CDR1/2/3 (α/β).", "link": "", "Citation_Count": "" }, { "id": "ml_dl-014", "name": "ERGO", "year": 2020, "method": "RNN", "methodology": "RNN", "input": "CDR3β; peptide", "data_source": "McPAS-TCR, VDJdb", "prediction_task": "TCR–peptide binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[26]", "ref": "[26]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide binding prediction Method: RNN. Input: CDR3β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-015", "name": "dGraph", "year": 2021, "method": "GNN", "methodology": "GNN", "input": "Unaligned FASTA sequences", "data_source": "-", "prediction_task": "Sequence similarity analysis and clustering", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[27]", "ref": "[27]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "Sequence similarity analysis and clustering Method: GNN. Input: Unaligned FASTA sequences.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-016", "name": "pMTnet", "year": 2021, "method": "RNN", "methodology": "RNN", "input": "CDR3β; pMHC", "data_source": "VDJdb, in-house datasets", "prediction_task": "TCR–pMHC binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[28]", "ref": "[28]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC binding prediction Method: RNN. Input: CDR3β; pMHC.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-017", "name": "DeepTCR", "year": 2021, "method": "CNN", "methodology": "CNN", "input": "CDR3α/β", "data_source": "10x Genomics, McPAS-TCR", "prediction_task": "TCR repertoire analysis", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[29]", "ref": "[29]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR repertoire analysis Method: CNN. Input: CDR3α/β.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-018", "name": "TCRGP", "year": 2021, "method": "Gaussian process", "methodology": "Gaussian process", "input": "CDR1/2/2.5/3(α/β)", "data_source": "VDJdb, IEDB, McPAS-TCR", "prediction_task": "TCR–pMHC binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[30]", "ref": "[30]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC binding prediction Method: Gaussian process. Input: CDR1/2/2.5/3(α/β).", "link": "", "Citation_Count": "" }, { "id": "ml_dl-019", "name": "TITAN", "year": 2021, "method": "CNN + Dual-modal Attention Mechanism", "methodology": "CNN + Dual-modal Attention Mechanism", "input": "TCRβ; peptide", "data_source": "VDJdb, COVID-19 datasets", "prediction_task": "TCR–peptide specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[31]", "ref": "[31]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide specificity prediction Method: CNN + Dual-modal Attention Mechanism. Input: TCRβ; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-020", "name": "DLpTCR", "year": 2021, "method": "CNN", "methodology": "CNN", "input": "CDR3α/β; peptide", "data_source": "VDJdb, IEDB", "prediction_task": "TCR–peptide interaction prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[32]", "ref": "[32]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide interaction prediction Method: CNN. Input: CDR3α/β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-021", "name": "NetTCR-2.1", "year": 2022, "method": "CNN", "methodology": "CNN", "input": "CDR1/2/3 (α/β)", "data_source": "IEDB, VDJdb, McPAS-TCR, 10x Genomics", "prediction_task": "TCR–pMHC interaction prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[33]", "ref": "[33]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC interaction prediction Method: CNN. Input: CDR1/2/3 (α/β).", "link": "", "Citation_Count": "" }, { "id": "ml_dl-022", "name": "MINN-SA", "year": 2022, "method": "Sparse-attention multi-instance neural network", "methodology": "Sparse-attention multi-instance neural network", "input": "TCRβ", "data_source": "TCGA", "prediction_task": "Tumor-specific TCR identification", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[34]", "ref": "[34]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "Tumor-specific TCR identification Method: Sparse-attention multi-instance neural network. Input: TCRβ.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-023", "name": "ColabFold", "year": 2022, "method": "Deep Evoformer", "methodology": "Deep Evoformer", "input": "TCRα/β; MHC; peptide", "data_source": "PDB70, BFD, MGnify, UniRef90, ColabFoldDB", "prediction_task": "TCR–pMHC structural prediction", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[20]", "ref": "[20]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC structural prediction Method: Deep Evoformer. Input: TCRα/β; MHC; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-024", "name": "AttnTAP", "year": 2022, "method": "RNN+ attention", "methodology": "RNN+ attention", "input": "CDR3β; peptide", "data_source": "VDJdb, IEDB, McPAS-TCR", "prediction_task": "TCR–peptide binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[35]", "ref": "[35]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide binding prediction Method: RNN+ attention. Input: CDR3β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-025", "name": "ATMTCR", "year": 2022, "method": "Transformer", "methodology": "Transformer", "input": "CDR3β", "data_source": "TCRdb", "prediction_task": "TCR–pMHC specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[36]", "ref": "[36]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC specificity prediction Method: Transformer. Input: CDR3β.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-026", "name": "DiffRBM", "year": 2023, "method": "Restricted Boltzmann machine + transfer learning", "methodology": "Restricted Boltzmann machine + transfer learning", "input": "TCRβ; peptide;", "data_source": "IEDB, VDJdb", "prediction_task": "Antigen immunogenicity and TCR specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[37]", "ref": "[37]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "Antigen immunogenicity and TCR specificity prediction Method: Restricted Boltzmann machine + transfer learning. Input: TCRβ; peptide;.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-027", "name": "SC-AIR-BERT", "year": 2023, "method": "BERT", "methodology": "BERT", "input": "TCRα/β", "data_source": "VDJdb, 10x PBMC, huARdb, CoV-AbDab", "prediction_task": "TCR–pMHC specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[38]", "ref": "[38]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC specificity prediction Method: BERT. Input: TCRα/β.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-028", "name": "TABR-BERT", "year": 2023, "method": "BERT", "methodology": "BERT", "input": "CDR3β; MHC; peptide", "data_source": "IEDB, VDJdb, McPAS-TCR", "prediction_task": "TCR–pMHC binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[39]", "ref": "[39]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC binding prediction Method: BERT. Input: CDR3β; MHC; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-029", "name": "epiTCR", "year": 2023, "method": "Random forest", "methodology": "Random forest", "input": "CDR3β; peptide", "data_source": "IEDB, TBAdb, VDJdb, McPAS-TCR, 10x", "prediction_task": "TCR–peptide binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[40]", "ref": "[40]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide binding prediction Method: Random forest. Input: CDR3β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-030", "name": "TEIM-Res", "year": 2023, "method": "CNN + few-shot pretraining", "methodology": "CNN + few-shot pretraining", "input": "CDR3β; pMHC", "data_source": "VDJdb, McPAS-TCR, COVID-19 datasets, IEDB", "prediction_task": "Residue-level TCR–peptide interaction prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[41]", "ref": "[41]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "Residue-level TCR–peptide interaction prediction Method: CNN + few-shot pretraining. Input: CDR3β; pMHC.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-031", "name": "MITNet", "year": 2023, "method": "CNN + Transformer", "methodology": "CNN + Transformer", "input": "CDR3β; TCRβV/J; peptide", "data_source": "IEDB, VDJdb, McPAS-TCR", "prediction_task": "TCR–peptide binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[42]", "ref": "[42]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide binding prediction Method: CNN + Transformer. Input: CDR3β; TCRβV/J; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-032", "name": "TPBTE", "year": 2023, "method": "CNN + Transformer", "methodology": "CNN + Transformer", "input": "CDR3β; peptide", "data_source": "IEDB, VDJdb, McPAS-TCR", "prediction_task": "TCR–peptide binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[43]", "ref": "[43]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide binding prediction Method: CNN + Transformer. Input: CDR3β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-033", "name": "Pan-pep", "year": 2023, "method": "Meta-learning + neural Turing machine + attention + CNN", "methodology": "Meta-learning + neural Turing machine + attention + CNN", "input": "CDR3β; peptide", "data_source": "IEDB, VDJdb, PIRD, McPAS-TCR", "prediction_task": "Pan-peptide TCR binding prediction", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[44]", "ref": "[44]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "Pan-peptide TCR binding prediction Method: Meta-learning + neural Turing machine + attention + CNN. Input: CDR3β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-034", "name": "NetTCR-2.2", "year": 2024, "method": "CNN", "methodology": "CNN", "input": "CDR1/2/3 (α/β); peptide", "data_source": "IEDB, VDJdb", "prediction_task": "TCR specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[45]", "ref": "[45]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR specificity prediction Method: CNN. Input: CDR1/2/3 (α/β); peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-035", "name": "TouCAN", "year": 2024, "method": "CNN + contrastive learning", "methodology": "CNN + contrastive learning", "input": "CDR1/2/3 (α/β)", "data_source": "VDJdb, IEDB, McPAS-TCR", "prediction_task": "TCR clustering and antigen-specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[46]", "ref": "[46]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR clustering and antigen-specificity prediction Method: CNN + contrastive learning. Input: CDR1/2/3 (α/β).", "link": "", "Citation_Count": "" }, { "id": "ml_dl-036", "name": "TEPCAM", "year": 2024, "method": "CNN + attention", "methodology": "CNN + attention", "input": "CDR3β; peptide", "data_source": "IEDB, VDJdb, McPAS-TCR", "prediction_task": "TCR–peptide specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[47]", "ref": "[47]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide specificity prediction Method: CNN + attention. Input: CDR3β; peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-037", "name": "Tspred", "year": 2024, "method": "CNN + attention", "methodology": "CNN + attention", "input": "CDR1/2/3 (α/β); peptide", "data_source": "VDJdb, IEDB, 10x, in-house datasets", "prediction_task": "TCR–peptide interaction prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[48]", "ref": "[48]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–peptide interaction prediction Method: CNN + attention. Input: CDR1/2/3 (α/β); peptide.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-038", "name": "BATMAN", "year": 2024, "method": "Bayesian", "methodology": "Bayesian", "input": "CDR3β; pMHC", "data_source": "IEDB, VDJdb, McPAS-TCR", "prediction_task": "Prediction of TCR response to single-amino-acid mutants", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[49]", "ref": "[49]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "Prediction of TCR response to single-amino-acid mutants Method: Bayesian. Input: CDR3β; pMHC.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-039", "name": "LightCTL", "year": 2025, "method": "CNN + contrastive learning", "methodology": "CNN + contrastive learning", "input": "CDR3β; peptide; MHC", "data_source": "McPAS-TCR, YFV, PIRD, VDJdb, IEDB, 10x, COVID-19, in-house datasets", "prediction_task": "TCR–pMHC specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[50]", "ref": "[50]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC specificity prediction Method: CNN + contrastive learning. Input: CDR3β; peptide; MHC.", "link": "", "Citation_Count": "" }, { "id": "ml_dl-040", "name": "SageTCR", "year": 2025, "method": "GNN", "methodology": "GNN", "input": "Residue- and atom-level TCR–pMHC structures", "data_source": "PDB; augmented/modeled structural data", "prediction_task": "TCR–pMHC binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[51]", "ref": "[51]", "category": "ml_dl", "category_label": "Machine learning / deep learning specificity models", "category_description": "ML/DL methods for TCR-peptide or TCR-pMHC specificity prediction, representation learning, and ligand discovery.", "supplemental_table": "Table S4", "brief_description": "TCR–pMHC binding prediction Method: GNN. Input: Residue- and atom-level TCR–pMHC structures.", "link": "", "Citation_Count": "" }, { "id": "multimodal-041", "name": "iCrossR", "year": 2017, "method": "CR indices and tissue profiles", "methodology": "CR indices and tissue profiles", "input": "Peptide; MHC allele", "data_source": "PaxDB; NCBI RefSeq; NetMHC", "prediction_task": "Cross-reactivity index estimation and tissue risk mapping", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[52]", "ref": "[52]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "Cross-reactivity index estimation and tissue risk mapping Method: CR indices and tissue profiles. Input: Peptide; MHC allele.", "link": "", "Citation_Count": "" }, { "id": "multimodal-042", "name": "sCRAP", "year": 2021, "method": "Rule-based scoring + conventional prediction tools", "methodology": "Rule-based scoring + conventional prediction tools", "input": "Peptide; MHC allele", "data_source": "Immunopeptidome databases of healthy tissues", "prediction_task": "pMHC cross-reactivity risk prediction", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[53]", "ref": "[53]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "pMHC cross-reactivity risk prediction Method: Rule-based scoring + conventional prediction tools. Input: Peptide; MHC allele.", "link": "", "Citation_Count": "" }, { "id": "multimodal-043", "name": "tessa", "year": 2021, "method": "Bayesian modeling", "methodology": "Bayesian modeling", "input": "CDR3β; single-cell RNA-seq data", "data_source": "TCGA; IEDB; McPAS-TCR; in-house datasets", "prediction_task": "Estimation of TCR effects on T-cell phenotypes", "cross_reactivity_use": "Indirect", "CR_Focused": "Indirect", "reference": "[54]", "ref": "[54]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "Estimation of TCR effects on T-cell phenotypes Method: Bayesian modeling. Input: CDR3β; single-cell RNA-seq data.", "link": "", "Citation_Count": "" }, { "id": "multimodal-044", "name": "TCR-Engine", "year": 2022, "method": "Experimental engineering + computational analysis", "methodology": "Experimental engineering + computational analysis", "input": "CDR3β; Peptide; MHC allele", "data_source": "In-house datasets", "prediction_task": "TCR engineering and specificity optimization", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[55]", "ref": "[55]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "TCR engineering and specificity optimization Method: Experimental engineering + computational analysis. Input: CDR3β; Peptide; MHC allele.", "link": "", "Citation_Count": "" }, { "id": "multimodal-045", "name": "CrossDome", "year": 2023, "method": "Multi-omics integration + statistical modeling", "methodology": "Multi-omics integration + statistical modeling", "input": "Peptide; HLA allele", "data_source": "HLA Ligand Atlas; HLAthena; IEDB; in-house datasets", "prediction_task": "T-cell cross-reactivity prediction", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[56]", "ref": "[56]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "T-cell cross-reactivity prediction Method: Multi-omics integration + statistical modeling. Input: Peptide; HLA allele.", "link": "", "Citation_Count": "" }, { "id": "multimodal-046", "name": "PepSim", "year": 2023, "method": "Structural and biochemical similarity analysis", "methodology": "Structural and biochemical similarity analysis", "input": "Peptide; pMHC complex structures", "data_source": "IEDB; CrossTope; in-house datasets", "prediction_task": "T-cell cross-reactivity prediction", "cross_reactivity_use": "Explicit", "CR_Focused": "Explicit", "reference": "[57]", "ref": "[57]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "T-cell cross-reactivity prediction Method: Structural and biochemical similarity analysis. Input: Peptide; pMHC complex structures.", "link": "", "Citation_Count": "" }, { "id": "multimodal-047", "name": "DeepAIR", "year": 2023, "method": "Transformer + CNN + RNN", "methodology": "Transformer + CNN + RNN", "input": "V(D)J gene usage; sequence features; optional structural information", "data_source": "10x Genomics; SARS-CoV-2 datasets; IEDB; TCRdb; in-house datasets", "prediction_task": "Immune receptor–peptide binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[58]", "ref": "[58]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "Immune receptor–peptide binding prediction Method: Transformer + CNN + RNN. Input: V(D)J gene usage; sequence features; optional structural information.", "link": "", "Citation_Count": "" }, { "id": "multimodal-048", "name": "Multimodal-AIR-BERT", "year": 2023, "method": "Transformer", "methodology": "Transformer", "input": "3-mer tokenized sequences; paired α/β or heavy/light chains", "data_source": "VDJdb; paired AIR-chain datasets", "prediction_task": "peptide-specificity prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[59]", "ref": "[59]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "peptide-specificity prediction Method: Transformer. Input: 3-mer tokenized sequences; paired α/β or heavy/light chains.", "link": "", "Citation_Count": "" }, { "id": "multimodal-049", "name": "pMTnet-omni", "year": 2023, "method": "Sequence + structural integration", "methodology": "Sequence + structural integration", "input": "TCR α/β; MHC allele; peptide", "data_source": "IEDB; GTEx; in-house datasets", "prediction_task": "cross-species TCR–pMHC binding prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[60]", "ref": "[60]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "cross-species TCR–pMHC binding prediction Method: Sequence + structural integration. Input: TCR α/β; MHC allele; peptide.", "link": "", "Citation_Count": "" }, { "id": "multimodal-050", "name": "TranspMHC", "year": 2024, "method": "Transformer", "methodology": "Transformer", "input": "MHC allele; peptide", "data_source": "IEDB; VDJdb; McPAS-TCR; in-house datasets", "prediction_task": "pMHC binding and TCR recognition prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[61]", "ref": "[61]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "pMHC binding and TCR recognition prediction Method: Transformer. Input: MHC allele; peptide.", "link": "", "Citation_Count": "" }, { "id": "multimodal-051", "name": "TRAP", "year": 2025, "method": "Multi-head attention + contrastive learning", "methodology": "Multi-head attention + contrastive learning", "input": "CDR3β; pMHC structural information", "data_source": "VDJdb; McPAS-TCR; IEDB", "prediction_task": "TCR–pMHC binding probability prediction", "cross_reactivity_use": "Adapted", "CR_Focused": "Adapted", "reference": "[62]", "ref": "[62]", "category": "multimodal", "category_label": "Multimodal cross-reactivity risk and engineering", "category_description": "Multimodal, risk-assessment, and engineering frameworks for cross-reactivity prioritization or contextual interpretation.", "supplemental_table": "Table S5", "brief_description": "TCR–pMHC binding probability prediction Method: Multi-head attention + contrastive learning. Input: CDR3β; pMHC structural information.", "link": "", "Citation_Count": "" } ] }