NeoFox: annotating neoantigen candidates with neoantigen features

AI Model for Predicting Anti-PD1 Response in Melanoma Using Multi-Omics Biomarkers

BACKGROUND

Immune checkpoint inhibitors (ICIs) have demonstrated significantly improved clinical efficacy in a minority of patients with advanced melanoma, whereas non-responders potentially suffer from severe side effects and delays in other treatment options. Predicting the response to anti-PD1 treatment in melanoma remains a challenge because the current FDA-approved gold standard, the nonsynonymous tumor mutation burden (nsTMB), offers limited accuracy.

METHODS

In this study, we developed a multi-omics-based machine learning model that integrates genomic and transcriptomic biomarkers to predict the response to anti-PD1 treatment in patients with advanced melanoma. We employed least absolute shrinkage and selection operator (LASSO) regression with 49 biomarkers extracted from tumor-normal whole-exome and RNA sequencing as input features. The performance of the multi-omics AI model was thoroughly compared to that of nsTMB alone and to models that use only genomic or transcriptomic biomarkers.

RESULTS

We used publicly available DNA and RNA-seq datasets of melanoma patients for the training and validation of our model, forming a meta-cohort of 449 patients for which the outcome was recorded as a RECIST score. The model substantially improved the prediction of anti-PD1 outcomes compared to nsTMB alone, with an ROC AUC of 0.7 in the training set and an ROC AUC of 0.64 in the test set. Using SHAP values, we demonstrated the explainability of the model's predictions on a per-sample basis.

CONCLUSIONS

We demonstrated that models using only RNA-seq or multi-omics biomarkers outperformed nsTMB in predicting the response of melanoma patients to ICI. Furthermore, our machine learning approach improves clinical usability by providing explanations of its predictions on a per-patient basis. Our findings underscore the utility of multi-omics data for selecting patients for treatment with anti-PD1 drugs. However, to train clinical-grade AI models for routine applications, prospective studies collecting larger melanoma cohorts with consistent application of exome and RNA sequencing are required.

pVACview: an interactive visualization tool for efficient neoantigen prioritization and selection

BACKGROUND

Neoantigen-targeting therapies including personalized vaccines have shown promise in the treatment of cancers, particularly when used in combination with checkpoint blockade therapy. At least 100 clinical trials involving these therapies have been initiated globally. Accurate identification and prioritization of neoantigens is crucial for designing these trials, predicting treatment response, and understanding mechanisms of resistance. With the advent of massively parallel DNA and RNA sequencing technologies, it is now possible to computationally predict neoantigens based on patient-specific variant information. However, numerous factors must be considered when prioritizing neoantigens for use in personalized therapies. Complexities such as alternative transcript annotations, various binding, presentation and immunogenicity prediction algorithms, and variable peptide lengths/registers all potentially impact the neoantigen selection process. There has been a rapid development of computational tools that attempt to account for these complexities. While these tools generate numerous algorithmic predictions for neoantigen characterization, results from these pipelines are difficult to navigate and require extensive knowledge of the underlying tools for accurate interpretation. This often leads to over-simplification of pipeline outputs to make them tractable, for example, limiting prediction to a single RNA isoform or only summarizing the top ranked of many possible peptide candidates. In addition to variant detection, gene expression, and predicted peptide binding affinities, recent studies have also demonstrated the importance of mutation location, allele-specific anchor locations, and variation of T-cell response to long versus short peptides. Due to the intricate nature and number of salient neoantigen features, presenting all relevant information to facilitate candidate selection for downstream applications is a difficult challenge that current tools fail to address.

RESULTS

We have created pVACview, the first interactive tool designed to aid in the prioritization and selection of neoantigen candidates for personalized neoantigen therapies including cancer vaccines. pVACview has a user-friendly and intuitive interface where users can upload, explore, select, and export their neoantigen candidates. The tool allows users to visualize candidates at multiple levels of detail including variant, transcript, peptide, and algorithm prediction information.

CONCLUSIONS

pVACview will allow researchers to analyze and prioritize neoantigen candidates with greater efficiency and accuracy in basic and translational settings. The application is available as part of the pVACtools software at pvactools.org and as an online server at pvacview.org.

pVACview: an interactive visualization tool for efficient neoantigen prioritization and selection

Background

Neoantigen targeting therapies including personalized vaccines have shown promise in the treatment of cancers, particularly when used in combination with checkpoint blockade therapy. At least 100 clinical trials involving these therapies are underway globally. Accurate identification and prioritization of neoantigens is highly relevant to designing these trials, predicting treatment response, and understanding mechanisms of resistance. With the advent of massively parallel DNA and RNA sequencing technologies, it is now possible to computationally predict neoantigens based on patient-specific variant information. However, numerous factors must be considered when prioritizing neoantigens for use in personalized therapies. Complexities such as alternative transcript annotations, various binding, presentation and immunogenicity prediction algorithms, and variable peptide lengths/registers all potentially impact the neoantigen selection process. There has been a rapid development of computational tools that attempt to account for these complexities. While these tools generate numerous algorithmic predictions for neoantigen characterization, results from these pipelines are difficult to navigate and require extensive knowledge of the underlying tools for accurate interpretation. This often leads to over-simplification of pipeline outputs to make them tractable, for example limiting prediction to a single RNA isoform or only summarizing the top ranked of many possible peptide candidates. In addition to variant detection, gene expression and predicted peptide binding affinities, recent studies have also demonstrated the importance of mutation location, allele-specific anchor locations, and variation of T-cell response to long versus short peptides. Due to the intricate nature and number of salient neoantigen features, presenting all relevant information to facilitate candidate selection for downstream applications is a difficult challenge that current tools fail to address.

Results

We have created pVACview, the first interactive tool designed to aid in the prioritization and selection of neoantigen candidates for personalized neoantigen therapies including cancer vaccines. pVACview has a user-friendly and intuitive interface where users can upload, explore, select and export their neoantigen candidates. The tool allows users to visualize candidates across three different levels, including variant, transcript and peptide information.

Conclusions

pVACview will allow researchers to analyze and prioritize neoantigen candidates with greater efficiency and accuracy in basic and translational settings The application is available as part of the pVACtools pipeline at pvactools.org and as an online server at pvacview.org.

Artificial intelligence and neoantigens: paving the path for precision cancer immunotherapy

Cancer immunotherapy has witnessed rapid advancement in recent years, with a particular focus on neoantigens as promising targets for personalized treatments. The convergence of immunogenomics, bioinformatics, and artificial intelligence (AI) has propelled the development of innovative neoantigen discovery tools and pipelines. These tools have revolutionized our ability to identify tumor-specific antigens, providing the foundation for precision cancer immunotherapy. AI-driven algorithms can process extensive amounts of data, identify patterns, and make predictions that were once challenging to achieve. However, the integration of AI comes with its own set of challenges, leaving space for further research. With particular focus on the computational approaches, in this article we have explored the current landscape of neoantigen prediction, the fundamental concepts behind, the challenges and their potential solutions providing a comprehensive overview of this rapidly evolving field.

Prediction of tumor-specific splicing from somatic mutations as a source of neoantigen candidates

Motivation

Neoantigens are promising targets for cancer immunotherapies and might arise from alternative splicing. However, detecting tumor-specific splicing is challenging because many non-canonical splice junctions identified in tumors also appear in healthy tissues. To increase tumor-specificity, we focused on splicing caused by somatic mutations as a source for neoantigen candidates in individual patients.

Results

We developed the tool splice2neo with multiple functionalities to integrate predicted splice effects from somatic mutations with splice junctions detected in tumor RNA-seq and to annotate the resulting transcript and peptide sequences. Additionally, we provide the tool EasyQuant for targeted RNA-seq read mapping to candidate splice junctions. Using a stringent detection rule, we predicted 1.7 splice junctions per patient as splice targets with a false discovery rate below 5% in a melanoma cohort. We confirmed tumor-specificity using independent, healthy tissue samples. Furthermore, using tumor-derived RNA, we confirmed individual exon-skipping events experimentally. Most target splice junctions encoded neoepitope candidates with predicted major histocompatibility complex (MHC)-I or MHC-II binding. Compared to neoepitope candidates from non-synonymous point mutations, the splicing-derived MHC-I neoepitope candidates had lower self-similarity to corresponding wild-type peptides. In conclusion, we demonstrate that identifying mutation-derived, tumor-specific splice junctions can lead to additional neoantigen candidates to expand the target repertoire for cancer immunotherapies.

Availability and implementation

The R package splice2neo and the python package EasyQuant are available at https://github.com/TRON-Bioinformatics/splice2neo and https://github.com/TRON-Bioinformatics/easyquant, respectively.

Virus-like particle-mediated delivery of structure-selected neoantigens demonstrates immunogenicity and antitumoral activity in mice

BACKGROUND

Neoantigens are patient- and tumor-specific peptides that arise from somatic mutations. They stand as promising targets for personalized therapeutic cancer vaccines. The identification process for neoantigens has evolved with the use of next-generation sequencing technologies and bioinformatic tools in tumor genomics. However, in-silico strategies for selecting immunogenic neoantigens still have very low accuracy rates, since they mainly focus on predicting peptide binding to Major Histocompatibility Complex (MHC) molecules, which is key but not the sole determinant for immunogenicity. Moreover, the therapeutic potential of neoantigen-based vaccines may be enhanced using an optimal delivery platform that elicits robust de novo immune responses.

METHODS

We developed a novel neoantigen selection pipeline based on existing software combined with a novel prediction method, the Neoantigen Optimization Algorithm (NOAH), which takes into account structural features of the peptide/MHC-I interaction, as well as the interaction between the peptide/MHC-I complex and the TCR, in its prediction strategy. Moreover, to maximize neoantigens' therapeutic potential, neoantigen-based vaccines should be manufactured in an optimal delivery platform that elicits robust de novo immune responses and bypasses central and peripheral tolerance.

RESULTS

We generated a highly immunogenic vaccine platform based on engineered HIV-1 Gag-based Virus-Like Particles (VLPs) expressing a high copy number of each in silico selected neoantigen. We tested different neoantigen-loaded VLPs (neoVLPs) in a B16-F10 melanoma mouse model to evaluate their capability to generate new immunogenic specificities. NeoVLPs were used in in vivo immunogenicity and tumor challenge experiments.

CONCLUSIONS

Our results indicate the relevance of incorporating other immunogenic determinants beyond the binding of neoantigens to MHC-I. Thus, neoVLPs loaded with neoantigens enhancing the interaction with the TCR can promote the generation of de novo antitumor-specific immune responses, resulting in a delay in tumor growth. Vaccination with the neoVLP platform is a robust alternative to current therapeutic vaccine approaches and a promising candidate for future personalized immunotherapy.

Multiple instance learning to predict immune checkpoint blockade efficacy using neoantigen candidates

Previous studies showed that the neoantigen candidate load is an imperfect predictor of immune checkpoint blockade (ICB) efficacy. Further studies provided evidence that the response to ICB is also affected by the qualitative properties of a few or even single candidates, limiting the predictive power based on candidate quantity alone. Here, we predict ICB efficacy based on neoantigen candidates and their neoantigen features in the context of the mutation type, using Multiple-Instance Learning via Embedded Instance Selection (MILES). Multiple instance learning is a type of supervised machine learning that classifies labeled bags that are formed by a set of unlabeled instances. MILES performed better compared with neoantigen candidate load alone for low-abundant fusion genes in renal cell carcinoma. Our findings suggest that MILES is an appropriate method to predict the efficacy of ICB therapy based on neoantigen candidates without requiring direct T cell response information.

Unraveling tumor specific neoantigen immunogenicity prediction: a comprehensive analysis

Introduction

Identification of tumor specific neoantigen (TSN) immunogenicity is crucial to develop peptide/mRNA based anti-tumoral vaccines and/or adoptive T-cell immunotherapies; thus, accurate in-silico classification/prioritization proves critical for cost-effective clinical applications. Several methods were proposed as TSNs immunogenicity predictors; however, comprehensive performance comparison is still lacking due to the absence of well documented and adequate TSN databases.

Methods

Here, by developing a new curated database having 199 TSNs with experimentally-validated MHC-I presentation and positive/negative immune response (ITSNdb), sixteen metrics were evaluated as immunogenicity predictors. In addition, by using a dataset emulating patient derived TSNs and immunotherapy cohorts containing predicted TSNs for tumor neoantigen burden (TNB) with outcome association, the metrics were evaluated as TSNs prioritizers and as immunotherapy response biomarkers.

Results

Our results show high performance variability among methods, highlighting the need for substantial improvement. Deep learning predictors were top ranked on ITSNdb but show discrepancy on validation databases. In overall, current predicted TNB did not outperform existing biomarkers.

Conclusion

Recommendations for their clinical application and the ITSNdb are presented to promote development and comparison of computational TSNs immunogenicity predictors.

Neodb: a comprehensive neoantigen database and discovery platform for cancer immunotherapy

Neoantigens derived from somatic deoxyribonucleic acid alterations are ideal cancer-specific targets. However, integrated platform for neoantigen discovery is urgently needed. Recently, many scattered experimental evidences suggest that some neoantigens are immunogenic, and comprehensive collection of these experimentally validated neoantigens is still lacking. Here, we have integrated the commonly used tools in the current neoantigen discovery process to form a comprehensive web-based analysis platform. To identify experimental evidences supporting the immunogenicity of neoantigens, we performed comprehensive literature search and constructed the database. The collection of public neoantigens was obtained by using comprehensive features to filter the potential neoantigens from recurrent driver mutations. Importantly, we constructed a graph neural network (GNN) model (Immuno-GNN) using an attention mechanism to consider the spatial interactions between human leukocyte antigen and antigenic peptides for neoantigen immunogenicity prediction. The new easy-to-use R/Shiny web-based neoantigen database and discovery platform, Neodb, contains currently the largest number of experimentally validated neoantigens. In addition to validated neoantigen, Neodb also includes three additional modules for facilitating neoantigen prediction and analysis, including 'Tools' module (comprehensive neoantigen prediction tools); 'Driver-Neo' module (collection of public neoantigens derived from recurrent mutations) and 'Immuno-GNN' module (a novel immunogenicity prediction tool based on a GNN). Immuno-GNN shows improved performance compared with known methods and also represents the first application of GNN model in neoantigen immunogenicity prediction. The construction of Neodb will facilitate the study of neoantigen immunogenicity and the clinical application of neoantigen-based cancer immunotherapy. Database URL https://liuxslab.com/Neodb/.

Seq2Neo: A Comprehensive Pipeline for Cancer Neoantigen Immunogenicity Prediction

Neoantigens derived from somatic DNA alterations are ideal cancer-specific targets. In recent years, the combination therapy of PD-1/PD-L1 blockers and neoantigen vaccines has shown clinical efficacy in original PD-1/PD-L1 blocker non-responders. However, not all somatic DNA mutations result in immunogenicity among cancer cells and efficient tools to predict the immunogenicity of neoepitopes are still urgently needed. Here, we present the Seq2Neo pipeline, which provides a one-stop solution for neoepitope feature prediction using raw sequencing data. Neoantigens derived from different types of genome DNA alterations, including point mutations, insertion deletions and gene fusions, are all supported. Importantly, a convolutional neural network (CNN)-based model was trained to predict the immunogenicity of neoepitopes and this model showed an improved performance compared to the currently available tools in immunogenicity prediction using independent datasets. We anticipate that the Seq2Neo pipeline could become a useful tool in the prediction of neoantigen immunogenicity and cancer immunotherapy. Seq2Neo is open-source software under an academic free license (AFL) v3.0 and is freely available at Github.