On TCR binding predictors failing to generalize to unseen peptides
Open accessThe DOI matches its Crossref record; 6 of the 6 references it cites were found in a registry or an index.
Abstract
Several recent studies investigate TCR-peptide/-pMHC binding prediction using machine learning or deep learning approaches. Many of these methods achieve impressive results on test sets, which include peptide sequences that are also included in the training set. In this work, we investigate how state-of-the-art deep learning models for TCR-peptide/-pMHC binding prediction generalize to unseen peptides. We create a dataset including positive samples from IEDB, VDJdb, McPAS-TCR, and the MIRA set, as well as negative samples from both randomization and 10X Genomics assays. We name this collection of samples TChard. We propose the hard split, a simple heuristic for training/test split, which ensures that test samples exclusively present peptides that do not belong to the training set. We investigate the effect of different training/test splitting techniques on the models’ test performance, as well as the effect of training and testing the models using mismatched negative samples generated randomly, in addition to the negative samples derived from assays. Our results show that modern deep learning methods fail to generalize to unseen peptides. We provide an explanation why this happens and verify our hypothesis on the TChard dataset. We then conclude that robust prediction of TCR recognition is still far for being solved.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
The relationship between Precision-Recall and ROC curves
Davis, J., & Goadrich, M. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143874 · Cited by 6,559
Saito, T., & Rehmsmeier, M. (2015) · PLOS ONE · doi:10.1371/journal.pone.0118432 · Cited by 5,309
A Probabilistic Interpretation of Precision, Recall and F-Score, with Implication for Evaluation
Goutte, C., & Gaussier, E. (2005) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-31865-1_25 · Cited by 2,285 · cited 1×
“The area under the receiver operator characteristic (AUROC) curve (33, 34), the area under the precision–recall (AUPR) curve (35, 36), the F1 score (F1) (37), and precision, recall, and classification accuracy are computed on the test sets and averaged.”
Identifying specificity groups in the T cell receptor repertoire
Glanville, J., Huang, H., Nau, A., Hatton, O., Wagar, L. E., Rubelt, F., Ji, X., Han, A., Krams, S. M., Pettus, C., et al. (2017) · Nature · doi:10.1038/nature22976 · Cited by 1,026
Diversity and clonal selection in the human T-cell repertoire
Qi, Q., Liu, Y., Cheng, Y., Glanville, J., Zhang, D., Lee, J.-Y., Olshen, R. A., Weyand, C. M., Boyd, S. D., & Goronzy, J. J. (2014) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1409155111 · Cited by 755 · cited 1×
“These alleles can be recombined unboundedly, which results in a high TCR repertoire diversity, essential for a broad T cell-based immune response (10).”
Understanding subset diversity in T cell memory
Jameson, S. C., & Masopust, D. (2018) · Immunity · doi:10.1016/j.immuni.2018.02.010 · Cited by 523
Cited by
Neoantigens: promising targets for cancer therapy
Xie, N., Shen, G., Gao, W., Huang, Z., Huang, C., & Fu, L. (2023) · Signal Transduction and Targeted Therapy · doi:10.1038/s41392-022-01270-x · Cited by 1,020
Phage display enables machine learning discovery of cancer antigen–specific TCRs
Croce, G., Lani, R., Tardivon, D., Bobisse, S., de Tiani, M., Bragina, M., Perez, M. A. S., Michaux, J., Pak, H. S., Michel, A., et al. (2025) · Science Advances · doi:10.1126/sciadv.ads5589 · Cited by 23
Assessing the generalization capabilities of TCR binding predictors via peptide distance analysis
Castorina, L. V., Grazioli, F., Machart, P., Mösch, A., & Errica, F. (2025) · PLOS One · doi:10.1371/journal.pone.0324011 · Cited by 7
Self-citationAssessing the Generalization Capabilities of TCR Binding Predictors via Peptide Distance Analysis
Castorina, L. V., Grazioli, F., Machart, P., Mösch, A., & Errica, F. (2023) · doi:10.1101/2023.07.29.551100 · Cited by 7
Self-citation
Metadata
- Authors
- Grazioli, F., Mösch, A., Machart, P., Li, K., Alqassem, I., O’Donnell, T. J., & Min, M. R.
- Venue
- Frontiers in Immunology
- Publisher
- Frontiers Media SA
- Published
- 2022-10-21
- Volume
- 13
- Pages
- 1014256-1014256
- ISSN
- 1664-3224 (ISSN-L)
- OpenAlex
- W4307166771
- Cited by
- 80
- References
- 6
- FWCI
- 4.49
- Type
- journal article
- Language
- en
- Open access
- gold
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.