Citation:

Ives, C., I. Campia, R. Wang, C. Wittwehr, AND S. Edwards. Computable AOPs Based on Biological Ontologies. NC SOT Annual Meeting, RTP, NC, October 25, 2016.

Impact/Purpose:

This work describes efforts to expand the AOP ontology developed as part of the international AOP-KB effort via the incorporation of existing biological ontologies. This work is essential for the long-term sustainability of AOP development and use. First, the expert-derived AOPs housed in the AOP-KB must be identifiable by controlled terminology to promote interoperability among the different modules of the knowledgebase and to facilitate the emergence of AOP networks. Second, having key events tied to biological ontologies is necessary for the incorporation of computationally predicted AOPs derived from data mining of toxicological databases. Finally, having the key events within an AOP tied to formal biological ontologies will promote direct use of the AOP information by computational methods and should facilitate the development of computational models describing the AOPs and AOP networks.

Description:

The Adverse Outcome Pathway (AOP) framework connects molecular perturbations with organism and population level endpoints used for regulatory decision-making by providing a conceptual construct of the mechanistic basis for toxicity. Describing AOPs via formal ontologies can enhance the reusability of Key Events (KEs), enable computational inference of relationships among AOPs, and facilitate the development of quantitative AOPs (qAOPs) backed by computational models. KEs representing over 150 AOPs from the AOP-Wiki (http://aopwiki.org/) were annotated using 22 publicly available controlled vocabularies and ontologies following a review of 66 different ontologies. Individual KEs were allowed to have one or more “event components” consisting of a biological process, object, and action with each term originating from one of the 22 biological ontologies. The biological context in which the KE occurs is described using cellular, organ, and species ontologies. The 22 ontologies were ranked based on the number of ontology classes or terms used. The Gene Ontology provided the majority of classes describing biological processes underlying KEs with 516 of 690 total classes. The Chemical Entities of Biological Interest ontology covered the most biological objects, with 172 of 565 total classes. Ontologies with infrequently used terms, as low as 1-3 classes across all KEs, will be re-evaluated and possibly removed if adequate terms can be identified in other ontologies. Following review by the original AOP authors, these terms will be included in the AOP-Wiki, and the selected ontologies will be used to drive author-selected tagging of new AOPs as they are entered. For 581 distinct KEs tagged, 762 of 1001 event components identified were reused or partially reused among KEs. Complete reuse of event components suggests that KEs should be merged; partial reuse of event components suggests looser associations among KEs that might result in cross-talk among AOPs without explicit sharing of KEs. Additionally, AOPs annotated with controlled vocabularies describing regulatory endpoints can better highlight the potential applications of AOPs for hazard and risk assessment across numerous legislations.[This is an abstract or a proposed presentation and does not necessarily reflect EPA policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.]

Record Details: