Abstract
Breast cancer is the most common cancer in females globally, accounting for 3660 diagnoses in New Zealand in 2020. Oestrogen receptor positive (ER+) is the most common subtype. ER+ breast cancer is in need of novel therapies to treat it, which requires new and improved models. Neoantigen vaccines are immunotherapies that target cancer-specific mutated proteins to boost antitumour T cell immunity. The SSM3 mammary tumour mouse model has been developed to research ER+ breast cancer. Matched SSM3 tumour and normal whole genome sequencing (WGS) and SSM3 RNA-seq data were used to predict immunogenic neoantigens, with the goal of designing an RNA vaccine to treat SSM3 tumours. The WGS data was aligned to the GRCm39 reference genome using BWA. VarScan2 detected somatic missense mutations, which pVACseq used to predict putative neoantigens. Kallisto aligned the RNA-seq data to the M33 reference transcriptome and calculated the transcript expression of these neoantigens. Neoantigens need to bind MHC molecules to activate T cells and the SSM3 H2 alleles encoding MHC are H2-Kb for MHC-I and H2-IAb for MHC-II. For neoantigens that were expressed, pVACseq predicted and ranked the MHC binding affinity of 25 H2-Kb neoantigens using NetMHCpan 4.1, and 17 H2-IAb neoantigens using NetMHCIIpan 4.1. NAP-CNB is designed for predicting MHC binding affinity and transcript expression in mice. NAP-CNB predicted 9 expressed H2-Kb neoantigens from raw SM3 RNA-seq data, that were different to those predicted by pVACseq. Top candidates were assessed for immunogenicity factors using the Immune Epitope Database, and DeepNeo. These results demonstrated the variation between neoantigen prediction tools, meaning immunogenicity needs to be validated in vitro before vaccination. This study also highlighted that neoantigen prediction tools need to be improved for mice, which are important for researching immunotherapies like vaccines.