Harlan Robins

TACO predicts TCR specificity against HLA-A*02:01-presented peptides using a statistical model, achieving AUROC 0.67 on held-out data and enabling efficient computational evaluation of TCR-peptide interactions.

HLAResolver improves HLA resolution in T-cell response data by integrating MIRA responses with NetMHCpan predictions, achieving higher sensitivity and fewer candidate HLAs.

This study identifies HLA allele-specific biases in TCR Vβ usage, showing germline-encoded preferences shape TCR-HLA compatibility and influence TCR-pHLA specificity.

The paper suggests that some autoimmune diseases may result from tissue defects rather than immune system dysregulation, based on new immune profiling data and existing studies.