An algorithm that combines machine learning ensemble modeling and network analysis to predict self-tolerant tumor-associated antigens for anti-cancer immunotherapy

Abstract

Tumor-associated antigens (TAAs) and their derived peptides constitute the chance to design off-the-shelf mainline or adjuvant anti-cancer immunotherapies for a broad array of patients. Here, we present a computational pipeline that selects and ranks candidate antigens in a multi-pronged approach and applied it to the case of uveal melanoma. In addition to antigen expression in the tumor target and in healthy tissues, we incorporated a network analysis-derived antigen indispensability index motivated by computational modeling results, and candidate immunogenicity predictions from a machine learning ensemble model on peptide physicochemical characteristics.