Dataset Information

NetTCR-2.0 enables accurate prediction of TCR-peptide binding by using paired TCRα and β sequence data.

ABSTRACT: Prediction of T-cell receptor (TCR) interactions with MHC-peptide complexes remains highly challenging. This challenge is primarily due to three dominant factors: data accuracy, data scarceness, and problem complexity. Here, we showcase that "shallow" convolutional neural network (CNN) architectures are adequate to deal with the problem complexity imposed by the length variations of TCRs. We demonstrate that current public bulk CDR3β-pMHC binding data overall is of low quality and that the development of accurate prediction models is contingent on paired α/β TCR sequence data corresponding to at least 150 distinct pairs for each investigated pMHC. In comparison, models trained on CDR3α or CDR3β data alone demonstrated a variable and pMHC specific relative performance drop. Together these findings support that T-cell specificity is predictable given the availability of accurate and sufficient paired TCR sequence data. NetTCR-2.0 is publicly available at https://services.healthtech.dtu.dk/service.php?NetTCR-2.0 .

SUBMITTER: Montemurro A

PROVIDER: S-EPMC8433451 | biostudies-literature | 2021 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

NetTCR-2.0 enables accurate prediction of TCR-peptide binding by using paired TCRα and β sequence data.

Montemurro Alessandro A Schuster Viktoria V Povlsen Helle Rus HR Bentzen Amalie Kai AK Jurtz Vanessa V Chronister William D WD Crinklaw Austin A Hadrup Sine R SR Winther Ole O Peters Bjoern B Jessen Leon Eyrich LE Nielsen Morten M

Communications biology 20210910 1

Prediction of T-cell receptor (TCR) interactions with MHC-peptide complexes remains highly challenging. This challenge is primarily due to three dominant factors: data accuracy, data scarceness, and problem complexity. Here, we showcase that "shallow" convolutional neural network (CNN) architectures are adequate to deal with the problem complexity imposed by the length variations of TCRs. We demonstrate that current public bulk CDR3β-pMHC binding data overall is of low quality and that the devel ...[more]

PMID: 34508155

Similar Datasets

Project description:The prediction of gene functions is crucial for a large number of different life science areas. Faster high throughput sequencing techniques generate more and larger datasets. The manual annotation by classical wet-lab experiments is not suitable for these large amounts of data. We showed earlier that the automatic sequence pattern-based BrEPS protocol, based on manually curated sequences, can be used for the prediction of enzymatic functions of genes. The growing sequence databases provide the opportunity for more reliable patterns, but are also a challenge for the implementation of automatic protocols. We reimplemented and optimized the BrEPS pattern generation to be applicable for larger datasets in an acceptable timescale. Primary improvement of the new BrEPS protocol is the enhanced data selection step. Manually curated annotations from Swiss-Prot are used as reliable source for function prediction of enzymes observed on protein level. The pool of sequences is extended by highly similar sequences from TrEMBL and SwissProt. This allows us to restrict the selection of Swiss-Prot entries, without losing the diversity of sequences needed to generate significant patterns. Additionally, a supporting pattern type was introduced by extending the patterns at semi-conserved positions with highly similar amino acids. Extended patterns have an increased complexity, increasing the chance to match more sequences, without losing the essential structural information of the pattern. To enhance the usability of the database, we introduced enzyme function prediction based on consensus EC numbers and IUBMB enzyme nomenclature. BrEPS is part of the Braunschweig Enzyme Database (BRENDA) and is available on a completely redesigned website and as download. The database can be downloaded and used with the BrEPScmd command line tool for large scale sequence analysis. The BrEPS website and downloads for the database creation tool, command line tool and database are freely accessible at http://breps.tu-bs.de.

Dataset Information

NetTCR-2.0 enables accurate prediction of TCR-peptide binding by using paired TCRα and β sequence data.

Publications

NetTCR-2.0 enables accurate prediction of TCR-peptide binding by using paired TCRα and β sequence data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets