Currently, neoantigens are often predicted using algorithms predominantly based on knowledge of the key peptide binding affinity difference between HLA alleles. Although HLA binding algorithms predict binding affinity of a peptide to HLA reasonably well, they do not predict processing and presentation of to the cell surface (i.e., the immunopeptidome). In fact, only 15%–20% of “predicted” peptide binders are processed or presented, and therefore contribute to the immunopeptidome. Erroneous predictions may be addressed with time-consuming and laborious experiments, such as mass-spectrometry (MS). However, in silico predictions may also prove to be very useful in prioritizing therapeutically relevant immunogenic peptides. Previous in silico studies that predict naturally processed and presented peptides to the cell surface have focused on only one of the many steps in the antigen processing and presentation pathway (such as TAP transport or proteasome cleavage, etc.). Additionally, previous antigen processing prediction tools have been trained and are therefore applicable to specific HLA alleles, making it challenging to make predictions for not so well-characterized alleles. Here, we outline a machine learning approach trained on MS elution data that predicts, in a pan-HLA manner, natural processing and presentation of neoantigens to the cell surface. The predictor is integrated with multiple immune parameters in a deep learning layer to predict neoantigens, and may be used for more accurate neoantigen predictions for any HLA allele, in both the class I and class II systems. Further, by analyzing previously published clinical data we illustrate that its application leads to a significantly improved identification of neoantigen targets for personalized cancer immunotherapy.

Citation Format: Trevor Clancy, Richard Stratford. A pan-HLA predictor of neoantigen processing and presentation to the tumor cell surface [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr B073.