Title Knowledge-based Systems in Immunology: Data Mining of Vaccine Targets from Viral Pathogens
Speaker Dr. Vladimir Brusic
Chair Tuan Pham

Abstract
Advances in genomics, proteomics, and bioinformatics have provided massive amounts of data that offer insights into the diversity of pathogens, their interaction with hosts, and molecular mechanisms that underlie responses to vaccines. Hundreds of thousands sequence variants are available for rapidly mutating pathogens such as HIV, influenza, or flaviviruses. Molecular databases focus primarily on storage, annotation and retrieval of these data and related information, but their infrastructure does not support detailed simultaneous analysis of large number of sequences and extraction of higher level knowledge. On the other hand, Knowledge-Based Systems (KBS) enable the extraction of high-level knowledge needed from problem solving and practical applications. The KBS use available data and specialized tools to extract knowledge, deploy knowledge representation to document knowledge, conceptual modeling to capture studied system properties, learning to acquire new knowledge, inference mechanisms to produce reasoning and explanation. The information-to-knowledge gap hampers practical applications such as development of diagnostics, prognostics, treatment, and prevention. For example, the development of vaccines against major viral pathogen requires the analysis of pathogen sequences and their variants, in combination with the knowledge about the pathogen, its antigens, and workings of the immune system. KBS are ideally suited for the automation of vaccine discovery and design. To bridge the gap between data and knowledge, we developed a framework for fast deployment of Web-accessible KBS. This framework enables rapid semi-automated data collection and integration, automated data storage and retrieval, and fast deployment of computational tools for in-depth analysis of various structural and functional properties associated with immune responses and vaccine development. The framework speeds up the immunological research and vaccine design by providing specialist databases hosting cleaned, well-annotated and structured data that are integrated into data mining pipeline for the discovery of new knowledge. The vaccine KBS provide a set of workflows that enable automated analysis of antigen diversity and their immunological potential through identification of potential T- or B-cell epitopes and their use for vaccine development. The visualization tools help display reports and results in an intuitive and easy-to-grasp format. Standardization of pathogen nomenclature enables automated quality control and quality assurance. The extended standardized nomenclature provides search keys for data mining, significantly increasing the speed and accuracy of data mining. By applying the KBS approach to influenza A, flaviviruses, and norovirus pathogen we increased the number of T-cell vaccine targets by an order of magnitude as compared to previously reported results. Furthermore, we were able to characterize and extract sequence patterns that define complete maps of cross-reactivities of neutralizing antibodies over hundred years of influenza data indicating cross-protection that can be afforded by individual vaccine formulations. Our results demonstrate the utility of KBS for immunology and vaccinology. KBS enable not only the discovery of new knowledge for immediate practical applications, but thanks to automation, they significantly reduce the time needed for the analysis.

Biography
Dr. Brusic is the Director of the Cancer Vaccine Center Bioinformatics Core at Dana-Farber Cancer Institute, and Principal Associate in Medicine Harvard Medical School. He is also a Professor of Computer Science at the Metropolitan College, Boston University. Previously, Dr. Brusic worked at the Walter and Eliza Hall Institute, Melbourne, Australia as bioinformatician, and at the Institute for Infocomm Research, Singapore where he was head of the Knowledge Discovery Department, and University of Queensland where he was Professor of Bioinformatics and Database Management.

Dr. Brusic holds BEng (Mechanical Engineering), MEng (Biomedical Engineering), MAppSci (Info Tech), MBA, and PhD degrees. He also holds a honorary doctorate (Doctor honoriscausa) from Semmelweis University, Budapest, Hungary. His research interests span the fields of biological databases, data mining, computational models of biological systems, simulation of molecular interactions, and biological discovery using simulation of laboratory experiments. The main focus of his work is computational modeling of complex biological systems such as immune system genome, and proteome.

He works on the development and application of computational methods in immunology and computer-aided development of vaccines. He has contributed a number of pioneering studies in computational immunology and immune-informatics. He introduced the use of several advanced machine learning techniques and developed novel algorithms for identification of T-cell epitopes and selection of vaccine targets. His current work includes development of computational solutions for the study of B-cell epitopes and integrative approaches to vaccine formulations. He is developing knowledge-based systems for biological data mining and knowledge discovery. Dr. Brusic is member of several editorial boards of scientific journals and has contributed to numerous scientific meetings.