HERVOminer: a sequence similarity-based approach for recognizing endogenous retrovirus origin of the peptidome.

Determining tumor-specific antigens (TSAs) derived from human endogenous retrovirus (HERV) regions is crucial because HERVs are a promising source of shared TSAs. However, systematic tools for accurately tracing peptide origins are limited by the high homology of HERV sequences, the complexity of quantifying multi-mapping reads, and a prior emphasis primarily on the long terminal repeat. To this end, we developed a sequence similarity-based approach (HERVOminer) to identify the HERV origins of peptides from our predefined HERV open-reading frame (ORFs) databases. HERVOminer compared sequence similarity between ORFs and candidate peptides, quantified their expression to determine the most expressed HERV fragment, and visualized the results. In our investigation of 15 colorectal cancer cohorts, candidate TSAs with HERV origins were found to be abundantly expressed and widely shared. Using ELISpot, we confirmed the immunogenicity of the candidate TSA. Furthermore, applying HERVOminer to HERV-derived peptides from preclinical and clinical studies, with quantification in the colorectal cancer samples, revealed broader genomic origins, with higher expression in tumor tissues compared to normal tissues. HERVOminer is implemented as an interactive web interface and a command-line package. HERVOminer can accelerate the development of novel therapeutic interventions such as off-the-shelf cancer vaccines.