Banerjee, Amitava, Pattinson, David, Wincek, Cornelia, Bunk, Paul, Chapin, Sarah, Navlakha, Saket, Meyer, Hannah (May 2024) BATMAN: Improved T cell receptor cross-reactivity prediction benchmarked on a comprehensive mutational scan database. In: UNSPECIFIED.
Abstract
Predicting T cell receptor (TCR) cross-reactivity is critical for understanding neoantigen immunogenicity and off-target toxicity of immunotherapies, but remains challenging due to (1) limited cross-reactivity assays and (2) positive example bias in databases. So TCR binding prediction softwares lack the sensitivity required to discriminate amongst single amino acid mutants of a TCR's cognate epitope. To address this problem, we curated a TCR activation database with an unprecendented coverage of complete single-amino-acid mutational scan assays of 14 immunogenic epitopes and 66 sequenced TCRs, yielding 10,750 TCR-peptide-major histocompatibility complex (pMHC) pairs. We also introduce BATMAN, “Bayesian inference of Activation of TCR by Mutant Antigens" (github.com/meyer-lab-cshl/BATMAN), which outperformed existing TCR-pMHC techniques in predicting variant effects on TCR activation and learned the biochemistry of TCR-pMHC interactions -- (1) binding motifs located near the middle of the peptide, (2) large effects of aromatic-non-aromatic AA substitutions (e.g., V-F) affecting side-chain interactions, and (3) detrimental effects of substituting out immunogenic hydrophobic residues (e.g., I and L). These results demand more experiments generating positive and negative TCR binding instances by similar peptides to constrain available TCR-pMHC models, which currently fail to predict peptide mutational effects, and thus should be used with caution in clinical settings.
| Item Type: | Conference or Workshop Item (Paper) |
|---|---|
| Subjects: | bioinformatics bioinformatics > computational biology |
| CSHL Authors: | |
| Communities: | CSHL labs > Beyaz lab CSHL labs > Meyer Lab CSHL labs > Navlakha lab |
| SWORD Depositor: | CSHL Elements |
| Depositing User: | CSHL Elements |
| Date: | 1 May 2024 |
| Date Deposited: | 19 Dec 2024 14:47 |
| Last Modified: | 19 Dec 2024 14:47 |
| URI: | https://repository.cshl.edu/id/eprint/41760 |
Actions (login required)
 |
Administrator's edit/view item |