Synopsis

• We have developed a new high-throughput, reproducible and sensitive immunopeptidomics platform that improves recovery and overall sensitivity for direct mass spectrometry based identification of neoantigens
• We have showed that incorporation of deconvoluted immunopeptidomics data in ligand prediction algorithms can improve their performance
• We have reported that by taking advantage of co-occurring HLA-I alleles one can rapidly and accurately identify HLA-I binding motifs and map them to their corresponding alleles without any a priori knowledge of HLA-I binding specificity. This novel and scalable approach uncovers new motifs for several alleles that had no known ligands
• In addition to providing huge training data to improve the HLA binding prediction, immunopeptidomics also captures other aspects of the natural in vivo presentation that significantly improve prediction of clinically relevant neoantigens

Synopsis

• Due to the high selectivity of the MHC molecules, major efforts have been dedicated to characterize their binding specificity and several in-silico methods have been developed to predict this event
• Overview of the recent advances in prediction methods for rational epitope discovery
• Show how the use of MS peptidomics data in can substantially enhance our ability to predict both MHC class I and MHC class II ligands and epitopes, and how the processing signal contained within MHC class II MS peptidome data can be used to improve predictive accuracy
• Demonstrate how peptide similarity to self serves as an important factor for predicting peptide immunogenicity
• Discuss how T cell receptor data can be used to refine our understanding of peptide immunogenicity, and present results suggesting that the cognate target of a T cell can be predicted from the sequence of T cell receptor
• Discussion on limitations of the state-of-the-art tools, and suggest solutions for how to move the field forward dealing with these limitations

Synopsis

• Current neoantigen discovery algorithms are not optimal to predict presentation to the cell surface
• Here, we outline a high-performing machine learning approach, trained on mass spectrometry data, that predicts naturally processed and presented antigens
• The predictor is integrated with several immune parameters, such as HLA binding, in a deep learning layer to predict bonafide neoantigens
• We illustrate its application to significantly improve the identification of neoantigen
  targets for personalized cancer immunotherapy

Synopsis

• Overview of factors for neoantigen prediction (Mutation detection, antigen presentation, antigen recognition)
• Establishing reliable mutation detection
• Strategies for neoantigen selection
• Mitigating the risks in selection processes
• Emerging factors

Synopsis

• How can you best predict the clinical relevance on your neoantigens?
• Use of in silico methods to enable the intelligent selection of mutations likely to result in high-affinity epitopes that bind to MHC molecules
• Analysis of immunoinformatics and computational tools for neoantigen prediction
• How can we harness machine learning to improve the efficiency of how neoantigens are validated?

Synopsis

• cfDNA neoantigens: dealing with tumor heterogeneity
• Improving neoantigen identification
• Elaborating TME, immuno-modulators and vaccine response biomarkers
• Overcoming poor sample quality and quantity for NGS sequencing
• Overcoming sequencing gaps that can harbor neoantigens
• Validation and regulatory issues on the way to commercialization

Synopsis

• Assessing the multidimensional challenge of identifying reactive neoantigens
• As sequence analysis tools and prediction algorithms promise optimal neoantigen calls, T cells status and tumour microenvironment can limit the number of truly reactive neoantigens in cancer patients
• Variant identification is a key limiting factor
• Neoantigen identification can be optimised by ranking epitopes based on combined prediction values
• Magnitude of TILs infiltration does not translate into optimal patient’s response
• TCR repertoire dictates response to neoantigen
• Neoantigens from driver mutations could also be exploited for augmented anti-tumor response

Synopsis

• Primary cancer data
• Metastatic cancer
• Normal tissue expression
• Functional characterisation

Synopsis

• The frequency of neoepitope reactive T cells is decreasing during TIL expansion but this can be circumvented in order to enrich TILs in neoepitope-specific T cells
• The neoepitope repertoire is discordant between circulating and tumor-infiltrating lymphocytes
• Neoepitope-specific T cells are functionally heterogeneous
• High-throughput strategies to validate neoepitopes and to isolate neoepitope-specific T cells are needed to bring mutanome-based immunotherapy into the clinic

Synopsis

• Specific content awaiting internal approval