search

GfellerLab / MixTCRpred

Predictor of TCR-epitope interactions
Other
16 stars 5 forks source link

pan-epitope-model #2

Closed ShadoWyyyy closed 4 months ago

ShadoWyyyy commented 4 months ago

Hi, brilliant work, I am wondering do users have access to pan-epitope-model for prediction of previous unseen allele-peptide-tcr binding prediction, do I miss the option or just not available yet, hope for response

GiancarloCroce commented 4 months ago

Hi, Thanks for your interest in our work. Currently, access to a pan-epitope model is not available. The pan-epitope model that we investigated in the paper (https://www.nature.com/articles/s41467-024-47461-8) is the "TransformerPredictor_AB_cdr123_with_epi" in the ./src/model.py file, but this is not supported by the MixTCRpred version on GitHub.

In the paper (Figure 3), we observed that extending predictions to unseen epitopes is challenging with the current amount of data. Successful predictions are possible only when the unseen epitope has very high sequence similarity with at least one epitope in the training set.

Best, Giancarlo

ShadoWyyyy commented 4 months ago

Hi, Thanks for your interest in our work. Currently, access to a pan-epitope model is not available. The pan-epitope model that we investigated in the paper (https://www.nature.com/articles/s41467-024-47461-8) is the "TransformerPredictor_AB_cdr123_with_epi" in the ./src/model.py file, but this is not supported by the MixTCRpred version on GitHub.

In the paper (Figure 3), we observed that extending predictions to unseen epitopes is challenging with the current amount of data. Successful predictions are possible only when the unseen epitope has very high sequence similarity with at least one epitope in the training set.

Best, Giancarlo

oh, I know, thanks for replying