Vladimir Brusic

Systematic identification of personal tumor-specific neoantigens in chronic lymphocytic leukemia.

Genome sequencing has revealed a large number of shared and personal somatic mutations across human cancers. In principle, any genetic alteration affecting a protein-coding region has the potential to generate mutated peptides that are presented by surface HLA class I proteins that might be recognized by cytotoxic T cells. To test this possibility, we implemented a streamlined approach for the prediction and validation of such neoantigens derived from individual tumors and presented by patient-specific HLA alleles. We applied our computational pipeline to 91 chronic lymphocytic leukemias (CLLs) that underwent whole-exome sequencing (WES). We predicted ∼22 mutated HLA-binding peptides per leukemia (derived from ∼16 missense mutations) and experimentally confirmed HLA binding for ∼55% of such peptides. Two CLL patients that achieved long-term remission following allogeneic hematopoietic stem cell transplantation were monitored for CD8(+) T-cell responses against predicted or confirmed HLA-binding peptides. Long-lived cytotoxic T-cell responses were detected against peptides generated from personal tumor mutations in ALMS1, C6ORF89, and FNDC3B presented on tumor cells. Finally, we applied our computational pipeline to WES data (N = 2488 samples) across 13 different cancer types and estimated dozens to thousands of predicted neoantigens per individual tumor, suggesting that neoantigens are frequent in most tumors.

Comprehensive analysis of cancer-associated somatic mutations in class I HLA genes

Detection of somatic mutations in human leukocyte antigen (HLA) genes using whole-exome sequencing (WES) is hampered by the high polymorphism of the HLA loci, which prevents alignment of sequencing reads to the human reference genome. We describe a computational pipeline that enables accurate inference of germline alleles of class I HLA-A, B and C genes and subsequent detection of mutations in these genes using the inferred alleles as a reference. Analysis of WES data from 7,930 pairs of tumor and healthy tissue from the same patient revealed 298 nonsilent HLA mutations in tumors from 266 patients. These 298 mutations are enriched for likely functional mutations, including putative loss-of-function events. Recurrence of mutations suggested that these hotspot sites were positively selected. Cancers with recurrent somatic HLA mutations were associated with upregulation of signatures of cytolytic activity characteristic of tumor infiltration by effector lymphocytes, supporting immune evasion by altered HLA function as a contributory mechanism in cancer.

HLA-binding properties of tumor neoepitopes in humans

Tumor neoantigens are common T-cell targets in humans with regressing or sometimes long-term stable disease. Fritsch and colleagues analyzed their predicted HLA-binding properties, and herein show that they provide important guidance for neoepitope selection for personalized cancer vaccines. Cancer genome sequencing has enabled the rapid identification of the complete repertoire of coding sequence mutations within a patients tumor and facilitated their use as personalized immunogens. Although a variety of techniques are available to assist in the selection of mutation-defined epitopes to be included within the tumor vaccine, the ability of the peptide to bind to patient MHC is a key gateway to peptide presentation. With advances in the accuracy of predictive algorithms for MHC class I binding, choosing epitopes on the basis of predicted affinity provides a rapid and unbiased approach to epitope prioritization. We show herein the retrospective application of a prediction algorithm to a large set of bona fide T cell–defined mutated human tumor antigens that induced immune responses, most of which were associated with tumor regression or long-term disease stability. The results support the application of this approach for epitope selection and reveal informative features of these naturally occurring epitopes to aid in epitope prioritization for use in tumor vaccines. Cancer Immunol Res; 2(6); 522–9. ©2014 AACR.

Computational resources for high-dimensional immune analysis from the Human Immunology Project Consortium

Computational resources for high-dimensional immune analysis from the Human Immunology Project Consortium

Pathway analysis and transcriptomics improve protein identification by shotgun proteomics from samples comprising small number of cells - a benchmarking study

BackgroundProteomics research is enabled with the high-throughput technologies, but our ability to identify expressed proteome is limited in small samples. The coverage and consistency of proteome expression are critical problems in proteomics. Here, we propose pathway analysis and combination of microproteomics and transcriptomics analyses to improve mass-spectrometry protein identification from small size samples.ResultsMultiple proteomics runs using MCF-7 cell line detected 4,957 expressed proteins. About 80% of expressed proteins were present in MCF-7 transcripts data; highly expressed transcripts are more likely to have expressed proteins. Approximately 1,000 proteins were detected in each run of the small sample proteomics. These proteins were mapped to gene symbols and compared with gene sets representing canonical pathways, more than 4,000 genes were extracted from the enriched gene sets. The identified canonical pathways were largely overlapping between individual runs. Of identified pathways 182 were shared between three individual small sample runs.ConclusionsCurrent technologies enable us to directly detect 10% of expressed proteomes from small sample comprising as few as 50 cells. We used knowledge-based approaches to elucidate the missing proteome that can be verified by targeted proteomics. This knowledge-based approach includes pathway analysis and combination of gene expression and protein expression data for target prioritization. Genes present in both the enriched gene sets (canonical pathways collection) and in small sample proteomics data correspond to approximately 50% of expressed proteomes in larger sample proteomics data. In addition, 90% of targets from canonical pathways were estimated to be expressed. The comparison of proteomics and transcriptomics data, suggests that highly expressed transcripts have high probability of protein expression. However, approximately 10% of expressed proteins could not be matched with the expressed transcripts.

HPVdb: a data mining system for knowledge discovery in human papillomavirus with applications in T cell immunology and vaccinology.

High-risk human papillomaviruses (HPVs) are the causes of many cancers, including cervical, anal, vulvar, vaginal, penile and oropharyngeal. To facilitate diagnosis, prognosis and characterization of these cancers, it is necessary to make full use of the immunological data on HPV available through publications, technical reports and databases. These data vary in granularity, quality and complexity. The extraction of knowledge from the vast amount of immunological data using data mining techniques remains a challenging task. To support integration of data and knowledge in virology and vaccinology, we developed a framework called KB-builder to streamline the development and deployment of web-accessible immunological knowledge systems. The framework consists of seven major functional modules, each facilitating a specific aspect of the knowledgebase construction process. Using KB-builder, we constructed the Human Papillomavirus T cell Antigen Database (HPVdb). It contains 2781 curated antigen entries of antigenic proteins derived from 18 genotypes of high-risk HPV and 18 genotypes of low-risk HPV. The HPVdb also catalogs 191 verified T cell epitopes and 45 verified human leukocyte antigen (HLA) ligands. Primary amino acid sequences of HPV antigens were collected and annotated from the UniProtKB. T cell epitopes and HLA ligands were collected from data mining of scientific literature and databases. The data were subject to extensive quality control (redundancy elimination, error detection and vocabulary consolidation). A set of computational tools for an in-depth analysis, such as sequence comparison using BLAST search, multiple alignments of antigens, classification of HPV types based on cancer risk, T cell epitope/HLA ligand visualization, T cell epitope/HLA ligand conservation analysis and sequence variability analysis, has been integrated within the HPVdb. Predicted Class I and Class II HLA binding peptides for 15 common HLA alleles are included in this database as putative targets. HPVdb is a knowledge-based system that integrates curated data and information with tailored analysis tools to facilitate data mining for HPV vaccinology and immunology. To our best knowledge, HPVdb is a unique data source providing a comprehensive list of HPV antigens and peptides. Database URL: http://cvc.dfci.harvard.edu/hpv/

Bioinformatics for cancer immunotherapy target discovery

The mechanisms of immune response to cancer have been studied extensively and great effort has been invested into harnessing the therapeutic potential of the immune system. Immunotherapies have seen significant advances in the past 20xa0years, but the full potential of protective and therapeutic cancer immunotherapies has yet to be fulfilled. The insufficient efficacy of existing treatments can be attributed to a number of biological and technical issues. In this review, we detail the current limitations of immunotherapy target selection and design, and review computational methods to streamline therapy target discovery in a bioinformatics analysis pipeline. We describe specialized bioinformatics tools and databases for three main bottlenecks in immunotherapy target discovery: the cataloging of potentially antigenic proteins, the identification of potential HLA binders, and the selection epitopes and co-targets for single-epitope and multi-epitope strategies. We provide examples of application to the well-known tumor antigen HER2 and suggest bioinformatics methods to ameliorate therapy resistance and ensure efficient and lasting control of tumors.

Large-Scale Analysis of B-Cell Epitopes on Influenza Virus Hemagglutinin – Implications for Cross-Reactivity of Neutralizing Antibodies

Influenza viruses continue to cause substantial morbidity and mortality worldwide. Fast gene mutation on surface proteins of influenza virus result in increasing resistance to current vaccines and available antiviral drugs. Broadly neutralizing antibodies (bnAbs) represent targets for prophylactic and therapeutic treatments of influenza. We performed a systematic bioinformatics study of cross-reactivity of neutralizing antibodies (nAbs) against influenza virus surface glycoprotein hemagglutinin (HA). This study utilized the available crystal structures of HA complexed with the antibodies for the analysis of tens of thousands of HA sequences. The detailed description of B-cell epitopes, measurement of epitope area similarity among different strains, and estimation of antibody neutralizing coverage provide insights into cross-reactivity status of existing nAbs against influenza virus. We have developed a method to assess the likely cross-reactivity potential of bnAbs for influenza strains, either newly emerged or existing. Our method catalogs influenza strains by a new concept named discontinuous peptide, and then provide assessment of cross-reactivity. Potentially cross-reactive strains are those that share 100% identity with experimentally verified neutralized strains. By cataloging influenza strains and their B-cell epitopes for known bnAbs, our method provides guidance for selection of representative strains for further experimental design. The knowledge of sequences, their B-cell epitopes, and differences between historical influenza strains, we enhance our preparedness and the ability to respond to the emerging pandemic threats.

Tumor antigens as proteogenomic biomarkers in invasive ductal carcinomas

BackgroundThe majority of genetic biomarkers for human cancers are defined by statistical screening of high-throughput genomics data. While a large number of genetic biomarkers have been proposed for diagnostic and prognostic applications, only a small number have been applied in the clinic. Similarly, the use of proteomics methods for the discovery of cancer biomarkers is increasing. The emerging field of proteogenomics seeks to enrich the value of genomics and proteomics approaches by studying the intersection of genomics and proteomics data. This task is challenging due to the complex nature of transcriptional and translation regulatory mechanisms and the disparities between genomic and proteomic data from the same samples. In this study, we have examined tumor antigens as potential biomarkers for breast cancer using genomics and proteomics data from previously reported laser capture microdissected ER+ tumor samples.ResultsWe applied proteogenomic analyses to study the genetic aberrations of 32 tumor antigens determined in the proteomic data. We found that tumor antigens that are aberrantly expressed at the genetic level and expressed at the protein level, are likely involved in perturbing pathways directly linked to the hallmarks of cancer. The results found by proteogenomic analysis of the 32 tumor antigens studied here, capture largely the same pathway irregularities as those elucidated from large-scale screening of genomics analyses, where several thousands of genes are often found to be perturbed.ConclusionTumor antigens are a group of proteins recognized by the cells of the immune system. Specifically, they are recognized in tumor cells where they are present in larger than usual amounts, or are physiochemically altered to a degree at which they no longer resemble native human proteins. This proteogenomic analysis of 32 tumor antigens suggests that tumor antigens have the potential to be highly specific biomarkers for different cancers.

An adaptive genetic algorithm for selection of blood-based biomarkers for prediction of Alzheimer's disease progression.

BackgroundAlzheimers disease is a multifactorial disorder that may be diagnosed earlier using a combination of tests rather than any single test. Search algorithms and optimization techniques in combination with model evaluation techniques have been used previously to perform the selection of suitable feature sets. Previously we successfully applied GA with LR to neuropsychological data contained within the The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging, to select cognitive tests for prediction of progression of AD. This research addresses an Adaptive Genetic Algorithm (AGA) in combination with LR for identifying the best biomarker combination for prediction of the progression to AD.ResultsThe model has been explored in terms of parameter optimization to predict conversion from healthy stage to AD with high accuracy. Several feature sets were selected - the resulting prediction moddels showed higher area under the ROC values (0.83-0.89). The results has shown consistency with some of the medical research reported in literature.ConclusionThe AGA has proven useful in selecting the best combination of biomarkers for prediction of AD progression. The algorithm presented here is generic and can be extended to other data sets generated in projects that seek to identify combination of biomarkers or other features that are predictive of disease onset or progression.