A transformer-based model to predict peptide–HLA class I binding and optimize mutated peptides for vaccine design

Human leukocyte antigen (HLA) can recognize and bind foreign peptides to present them to specialized immune cells, then initiate an immune response. Computational prediction of the peptide and HLA (pHLA) binding can speed up immunogenic peptide screening and facilitate vaccine design. However, there is a lack of an automatic program to optimize mutated peptides with higher affinity to the target HLA allele. Here, to fill this gap, we develop the TransMut framework—composed of TransPHLA for pHLA binding prediction and an automatically optimized mutated peptides (AOMP) program—which can be generalized to any binding and mutation task of biomolecules. First, TransPHLA is developed by constructing a transformer-based model to predict pHLA binding, which is superior to 14 previous methods on pHLA binding prediction and neoantigen and human papilloma virus vaccine identification. For vaccine design, the AOMP program is then developed by exploiting the attention scores generated by TransPHLA to automatically optimize mutated peptides with higher affinity to the target HLA allele and with high homology to the source peptide. The proposed framework may automatically generate potential peptide vaccines for experimentalists. The human leukocyte antigen (HLA) complex plays an important role in building an immune response, but it is hard to predict which peptides will bind to it. Chu et al. present a transformer-based approach to identify which peptides have a high binding affinity to HLA, a task that can also be translated to other binding problems.