A tool to predict TCR-peptide-MHC binding based on embeddings derived from ESM, a large protein language model.

The tool can be accessed from the given link TCR-ESM

The tool will take inputs as numpy (.npy) files for the embeddings derived from the sequences of the TCR, peptide and MHCs.

The required input files can be generated using the ESM1v (https://github.com/facebookresearch/esm)

Procedure to extract embeddings for bulk fasta files is available at https://github.com/facebookresearch/esm#bulk_fasta

ESM1v generated embeddings can be converted to .npy files by https://sparrow.dev/pytorch-numpy-conversion/

For further details visit https://github.com/dhanjal-lab

Developed by shashank@arizona.edu