Microbial profiling, neural network and semantic web: an integrated information system for human pathogen risk management, prevention and surveillance in food safety

Authors: Yan, Xianghe; PENG, YUN - University Of Maryland; MENG, JIANGHONG - University Of Maryland; RUZANTE, JULIANA - University Of Maryland; Fratamico, Pina; Huang, Lihan; Juneja, Vijay

Submitted to: International Conference on Predictive Modeling in Foods
Publication Type: Proceedings
Publication Acceptance Date: 8/20/2009
Publication Date: 9/8/2009
Citation: Yan, X., Peng, Y., Meng, J., Ruzante, J., Fratamico, P.M., Huang, L., Juneja, V.K. 2009. Microbial profiling, neural network and semantic web: an integrated information system for human pathogen risk management, prevention and surveillance in food safety. International Conference on Predictive Modeling in Foods. 1:11-12.

Interpretive Summary:

Technical Abstract: It is estimated that food-borne pathogens cause approximately 76 million cases of gastrointestinal illnesses, 325,000 hospitalizations, and 5,000 deaths in the United States annually. Genomic, proteomic, and metabolomic studies, particularly, genome sequencing projects are providing valuable information for understanding pathogen behavior/adaptation in food and food processing environments and for development of methods for rapid detection and typing/subtyping of pathogens. The USDA’s Pathogen Modeling Program (PMP), Combase (a relational database of predictive food microbiology information), and the online food safety risk analysis resource, Foodrisk.org, developed by the Joint Institute for Food Safety and Applied Nutrition (JIFSAN - a joint institute of the U.S. FDA and the University of Maryland) continue to attract attention from researchers from academia, the food industry, and government agencies. However, the consistency among semantically heterogeneous data resources, the awareness of knowledge gaps among research communities, government risk assessors/managers, and end users of the information, and a clear understanding of the molecular profiling of food-borne pathogens and their responses to environmental factors such as oxygen, pH, temperature, osmotic pressure, storage conditions during transportation and handling at retail or by consumers are lacking. The systematic integration of heterogeneous data resources, molecular profiling for pathogen identification and characterization, risk analyses, and other information related to environmental factors that impact pathogen virulence and the ability to cause food-borne disease are essential for successful disease prevention and pathogen surveillance and for risk management strategies for the food industry and government agencies (USDA, CDC, and FDA). The aim of this project is to develop a comprehensive web-based internet resource and central reporting system that can computationally collect publicly available information, including microbial risk analyses, predictive modeling data, regulatory data/policies, and experimentally-derived molecular signature data from public domains. This information will be compiled, analyzed, and presented through the development of neural network style learning and semantic web technology to gain a better understanding of transmission patterns and dynamics of food-borne pathogens and outbreaks. We propose to: a) computationally collect and compile publicly available information, including published molecular profiling data relevant to genomic, proteomic, and metabolomic analyses of food-borne pathogens from various public resources; b) develop an ontology library on food-borne pathogens and design automatic algorithms with formal inference and fuzzy and probabilistic reasoning to address the consistency and accuracy of distributed information resources (e.g., PubMed, NCBI, EMBL, and online genetic databases and information); c) integrate newly collected molecular profiling data, PMP, Combase, Foodrisk.org (http://www.foodrisk.org/), and other relevant information into a user-friendly “homogeneous” information system, which will be understandable and easily accessed by scientists in academia and by the food industry and government agencies; and d) develop a computational model in semantic web for greater adaptability and robustness, providing a richer learning environment. The long-term goal of this project is to develop a comprehensive food safety information reporting system that will serve as an efficient early warning system in outbreak investigations and for food-borne pathogen surveillance and control. Data sharing, analyses, and integration are difficult; therefore, this system will be the key to developing a real-time monitoring system, which will contribute toward risk management, prevention of food-borne illness, outbrea