FoodON: A Global Farm-to-Fork Food Ontology The Development of a Universal Food Vocabulary Emma Griffiths, Fiona Brinkman Damion Dooley, Will Hsiao Department of Molecular Biology and Biochemistry BC Public Health Laboratory Simon Fraser University Vancouver, Canada Greater Vancouver, Canada Robert Hoehndorf Pier Buttigieg Department of Computer Science Alfred-Wegener-Institut, Helmholtz-Zentrum für Polar- und King Abdullah University Meeresforschung Thuwal, Saudi Arabia Bremerhaven, Germany Abstract— Several resources and standards for indexing food have major impacts in many aspects of life, from cultural descriptors currently exist, but their content and interrelations are practices to purchasing and economics to health and safety. not semantically and logically coherent. Simultaneously, the need Many food vocabularies have been developed to meet local to represent knowledge about food is central to many fields needs, but differences in their content, structure and purpose including biomedicine and sustainable development. FoodON is a limit their global adoption. As such, these resources conflict new ontology built to interoperate with the OBO Library and to when information must be shared on a broader basis. A food represent entities which bear a “food role”. It encompasses ontology which seeks to harmonize the semantics behind these materials in natural ecosystems and food webs as well as human- vocabularies would standardize information transfer between centric categorization and handling of food. The latter will be the national and international organizations, and allow integration initial focus of the ontology, and we aim to develop semantics for food safety, food security, the agricultural and animal husbandry of diverse data entities. practices linked to food production, culinary, nutritional and II. DEVELOPMENT OF THE FOODON FOOD ONTOLOGY chemical ingredients and processes. The scope of FoodON is ambitious and will require input from multiple domains. A. Potential Use Cases for a Global Food Ontology FoodON will import or map to material in existing ontologies and standards and will create content to cover gaps in the Initial interviews with domain experts (public health agencies, representation of food-related products and processes. As a robust food allergy strategy committee members, agriculture and agri- food ontology can only be created by consensus and wide adoption, food specialists, US Food and Drug Administration officials we are currently forming an international consortium to build etc) have suggested a number of use cases for which a global partnerships, solicit domain expertise, and gather use cases to food ontology will have immediate and considerable impact. guide the ontology’s development. The products of this work are These use cases include food safety (source attribution during being applied to research and clinical datasets such as those outbreaks, contamination traceback, risk assessment), food associated with the Canadian Healthy Infant Longitudinal Development (CHILD) study which examines the causal factors of security (production, storage, processing, distribution, asthma and allergy development in children, and the Integrated consumption), food allergy, chemical exposure and nutritional Rapid Infectious Disease Analysis (IRIDA) platform for genomic assessment, food regulation, trade (import/export/economics), epidemiology and foodborne outbreak investigation. cultural practices, cooking and culinary use, food webs/food cycles and animal husbandry. An example of how a food Keywords—food ontology, food safety, food security, food ontology would have immediate impact is in resolving allergy, nutrition discrepancies arising from food synonyms (e.g. cold cuts vs I. INTRODUCTION deli meats). Such discrepancies can obscure query results and have important consequences in real-time investigations. We All living things require food to survive. While some will use these use cases to guide development and test the organisms can utilize sunlight and chemicals to synthesize their competencies of FoodON as it develops. own nutrients, all animals require exogenous food sources. Food is such a critical resource for life that a substantial proportion of B. Harmonization of Existing Food Vocabularies time and effort in an animal’s life is spent procuring it. Humans Several resources and standards for indexing food descriptors have built intricate systems for the production, distribution, currently exist. The Codex Alimentarius is a collection of acquisition, regulation, treatment, consumption and disposal of internationally recognized standards, codes of practice, food. The relationships between humans and food are complex, guidelines, and other recommendations relating to foods, food and have real consequences for policy, security, education and production, and food safety commissioned by the United science. The ways in which we communicate about food can This work was supported by Genome Canada and Genome BC. Nations Food and Agriculture Organization III. FORMATION OF THE FOODON CONSORTIUM (http://www.fao.org/fao-who-codexalimentarius/codex- The creation of such a Food Ontology is a large undertaking home/en/). LanguaL, or the Langua aLimentaria [1], was first and is much more tractable through an open-member consortium developed in the late seventies by USFDA’s Centre for Food framework. As harmonization of the food ontology can only be Safety and Applied Nutrition, and is now hosted by a achieved by consensus and wide adoption, we are currently consultancy, Danish Food Informatics, on behalf of various forming an international consortium to build partnerships and European partners. LanguaL provides 14 main facets, or solicit domain expertise. An open-membership consortium will hierarchies of descriptive terms that have been used to describe be key to developing a sustainable and extensible FoodON. around 35,000 foods. It has been internationalized with Priorities of the consortium include term contributions, equivalent terms in Czech, Danish, English, French, German, integration of regionally-specific food vocabulary and strategies Italian, Portuguese, Spanish and Hungarian. Europe is also for international uptake. Subsets of FoodON will be utilized by home to FoodEx [2], a food dictionary constructed by the the Integrated Rapid Infectious Disease Analysis (IRIDA) European Food Safety Authority (EFSA). The latest version platform which supports real-time foodborne outbreak (Food Ex2) provides a comprehensive classification of terms investigation, as part of the Genomic Epidemiology Application and is designed to facilitate food exposure assessment. Other Ontology (GenEpiO, www.irida.ca). FoodON vocabulary will related resources include the USDA National Nutrient Database also be useful for standardizing food descriptions within the for Standard Reference [3], and the Health Canada Canadian Healthy Infant Longitudinal Development (CHILD) study, which examines the causal factors of asthma and allergy Compendium of Analytical Methods, designed to foster development in children. Interested participants should contact compliance of the food industry with standards and guidelines the authors for further details. relative to microbiological and extraneous material in foods [4]. Ontological representation of food products is somewhat REFERENCES scattered and not comparably comprehensive. For example, [1] Ireland JD & Møller A (2010). LanguaL food description: a following requests from its users due to the lack of an learning process. Eur J Clin Nutr. 64 Suppl 3:S44-8. independent food ontology, the Environment Ontology (ENVO) contained representations of a range of food products [2] European Food Safety Authority (2015). The food [5]. The experimental FoodO ontology was designed to classification and description system FoodEx2 (revision 2). represent the FooDB database describing food items and EFSA supporting publication 2015:EN-804. 90 pp. chemical/nutritional composition (http://aber- [3] Ahuja JK, Moshfegh AJ, Holden JM & Harris E (2013). owl.net/ontology/FOODO). While these resources are certainly USDA food and nutrient databases provide the infrastructure for useful, they have only been developed for specific uses and food and nutrition research, policy, and practice. J Nutr. often include different facets and organizational principles. 143(2):241S-9S. These features limit interoperability and reduce the range of [4] Health Canada (2011). “Guidelines for the Relative application an integrative and semantically coherent food Validation of Indirect Qualitative Food Microbiological ontology would possess. Methods” in The Compendium of Analytical Methods (Vol 1). [5] Buttigieg PL, Morrison N, Smith B, Mungall CJ, Lewis SE C. FoodON Structure and Content & the ENVO Consortium (2013). The environment ontology: We are developing FoodON following the principles of the contextualising biological and biomedical entities. J Biomed OBO Foundry and Library [6]. Its upper levels will be Semantics. 4(1):43. concerned with food composition, pre-collection (farming or [6] Smith B, Ashburner M, Rosse C, Bard J, Bug W, Ceusters animal husbandry) environment, collection features (season, W, Goldberg LJ, Eilback K, Ireland A, Mungall CJ, The OBI age of plant or animal, harvesting technology), storage and Consortium, Leontis N, Rocca-Serra P, Ruttenberg A,Sansone distribution features as they pertain to food security, culinary S-A, Scheuermann RH, Shah N, Whetzel PL & Lewis S (2007). preparation and packaging, food safety factors, and The OBO Foundry: coordinated evolution of ontologies to consumption patterns. Food-relevant terms are being imported support biomedical data integration. Nat Biotechnol. 25(11): from existing ontologies including NCBI Taxonomy for source 1251-5. http://doi.org/10.1038/nbt1346 animal / plant / fungi / bacterial organism, Uberon [7] for food source part and type descriptors, ENVO for food environmental [7] Mungall CJ, Torniai C, Gkoutos GV, Lewis SE & Haendel MA (2012). Uberon, an integrative multi-species anatomy context, and CHEBI [8] for chemical constituents and ontology. Genome Biol. 13(1), R5. http://doi.org/10.1186/gb- contaminants. LanguaL, a traditional database, will be used for 2012-13-1-r5 food source common names, preservation methods, and a few other facets. The ontology is open and available on GitHub [8] Degtyarenko K, de Matos P, Ennis M, Hadtings J, Zbinden (https://github.com/FoodOntology) such that editors, each with M, McNaught A, Alcantara R, Darsow M, Guedj M & their own allotment of term identifiers, can effectively Ashburner M (2008). ChEBI: a database ontology for chemical contribute in parallel and the community can submit term entities of biological interest. Nucleic Acids Res. 36 (Database requests and feedback. issue), D344-50. http://doi.org/10.1093/nar/gkm791 This work was supported by Genome Canada and Genome BC.