Abstract:
Cancer immunotherapy is a treatment option that involves or uses components of a patient’s immune system. Today, it is heading towards becoming an integral part of treatment plans together with chemotherapy, surgery, and radiotherapy. Personalized epitope-based vaccines (EVs) serve as one strategy that is truly personalized. Each patient possesses a distinct immune system, and each tumor is unique, rendering the design of a potent vaccine challenging and dependent on the patient and the tumor. The potency of a vaccine is reliant on the ability of its constituent epitopes – short, immunogenic antigen fragments – to trigger an immune response. To assess this ability, one has to take into account the individuality of the immune system, among others conditioned by the variability of the human leukocyte antigen (HLA) gene cluster. Determining the HLA genotype with traditional experimental techniques can be time- and cost-intensive. We proposed a novel HLA genotyping algorithm based on integer linear programming that is independent of dedicated data generation for the sole purpose of HLA typing. On publicly available next-generation sequencing (NGS) data, our method outperformed previously published approaches. HLA binding is a prerequisite for T-cell recognition, and precise prediction algorithms exist. However, this information is not sufficient to assess the immunogenic potential of a peptide. To induce an immune response, reactive T-cell clones with receptors specific for a peptide-HLA complex have to be present. We suggested a method for the prediction of immunogenicity that includes peripheral tolerance models, based on gut microbiome data, in addition to central tolerance, previously shown to increase performance. The comparison to a previously published method suggests that the incorporation of gut microbiome data and HLA-binding stability estimates do not enhance prediction performance. High-throughput sequencing provides the basis for the design of personalized EVs. Through genome and transcriptome sequencing of tumor and matched non-malignant tissue samples, cancer-specific mutations can be identified, which can be further validated using other technologies such as mass spectrometry (MS). Multi-omics approaches can result in the acquisition of several hundreds of gigabytes of data. Handling and analysis of such data usually require data management solutions and high-performance computing (HPC) infrastructures. We developed the web-based platform qPortal for data-driven biomedical research that allows users to manage and analyze quantitative biological data intuitively. To emphasize the advantages of our data-driven approach with an integrated workflow system, we conducted a comparison to Galaxy. Building on qPortal, we implemented the web-based platform iVacPortal for the design of personalized EVs to facilitate data management and data analysis in such projects. Further, we applied the implemented methods through iVacPortal in two studies of two distinct cancer entities, indicating the added value of our platform for the assessment of personalized EV candidates and alternative targets for cancer immunotherapy.
Next-Generation Sequencing