We are excited to announce that IMMREP25 launches today (March 3rd) and will be running for three weeks until March 23rd. This initiative originates from the AIRR community and is supported by IEDB staff this year.
IMMREP is a TCR-pMHC specificity prediction challenge that ran in 2022, 2023 and is running again now in 2025. The challenge is hosted on Kaggle.
This year the focus is on prediction of TCR specificity to unseen pHLA. Adaptive Biotechnologies have generously provided a test dataset of paired chain alpha-beta TCRs for 20 peptides from common viral pathogens restricted to either HLA-A:02*01 or HLA-B:40*01. In total, there will be ~9,000 TCR-pHLA pairs to evaluate. This represents an unprecedented opportunity to benchmark prediction models against a uniform, blinded test set to quantify how well these models generalise.
Consider participating, or just follow along to learn more about the state-of-the-art in TCR specificity prediction!
The IEDB contains human TCR sequence data from >300 papers. Currently, our curated fields can contain values that are difficult to work with for bioinformaticians as they are author-reported, and may no longer reflect naming conventions. We are standardizing these values in the future so you'll just be able to use the Calculated fields, and are making efforts to improve the usability of our receptor data. We have made available a dataset of human alpha-beta TCR-pHLA pairs on Kaggle as a starting point for training data; this is described here.
Comments
0 comments
Article is closed for comments.