Adding evidence type representation to DIDEO Mathias Brochhausen William R. Hogan Department of Biomedical Informatics Department of Health Outcomes and Policy University of Arkansas for Medical Sciences University of Florida Little Rock, AR USA Gainesville, FL USA mbrochhausen@uams.edu Philip E. Empey Jodi Schneider, Richard D. Boyce Department of Pharmacy and Therapeutics Department of Biomedical Informatics University of Pittsburgh University of Pittsburgh Pittsburgh, PA USA Pittsburgh, PA USA Abstract—In this poster we present novel development and DDIs and PDDIs. An important use case for the new extension of the Drug-drug Interaction and Drug-drug representation is to automatically categorize evidence items Interaction Evidence Ontology (DIDEO). We demonstrate how into multilevel taxonomy of evidence types. We plan for reasoning over this extension of DIDEO can a) automatically curators of DDI and PDDI information to use a web-based data create a multi-level hierarchy of evidence types from descriptions entry form to enter information about a scientific observation of the underlying scientific observations and b) automatically that the particular evidence item is about (e.g. an experiment, a subsume individual evidence items under the correct evidence clinical study, a case report, etc.). Examples of the aspects of type. Thus DIDEO will enable evidence items added manually by scientific observations relevant to our use case include among curators to be automatically categorized into a drug-drug others: group randomization, targeting pharmacokinetics, interaction framework with precision and minimal effort from curators. As with all previous DIDEO development this extension number of drugs involved, enzymes involved, inclusion of is consistent with OBO Foundry principles. antibodies, etc. Based on information about these aspects we want to enable automatic categorization of our evidence items Keywords—drug-drug interaction; potential drug-drug into the DIKB evidence type taxonomy [9]. The top level of interaction; evidence types; biomedical ontologies this evidence taxonomy is: I. INTRODUCTION  Statements of various kinds The Drug-drug Interaction and Drug-drug Interaction  Metabolic enzyme identification experiments Evidence Ontology (DIDEO) is an ontology aimed at  Metabolic enzyme inhibition experiments representing drug-drug interactions, potential drug-drug interactions and the underlying phenomena from physiology,  Transport protein identification experiments anatomy, pharmacology and laboratory science. The goal in creating DIDEO is to provide a realism-based, semantically  Transport protein inhibition experiments rich, and logically consistent OWL representation for the Drug  Prospective clinical studies Interaction Knowledge Base (DIKB) [1,2]. DIDEO is based on Basic Formal Ontology [3] and is compliant with the OBO  Non-randomized studies and case reports Foundry [4] principles [5]. It is coded in Web Ontology Language (OWL2) [6] and is freely accessible from  Observational studies http://purl.obolibrary.org/obo/dideo.owl. II. METHODS A key achievement of the initial version of DIDEO [7] was The key strategy for achieving automatic categorization of to establish a clear distinction between drug-drug interactions evidence is to use a) necessary and sufficient conditions of or DDIs (biological processes) and potential drug-drug evidence types and b) property assertions for evidence items interactions or PDDIs (information content entities) based on and the related scientific observations. Fig. 1 shows the classes the paradigm of ontological realism [8]. This deliberate and relations used to create the necessary and sufficient axiom separation of representations of physiological processes and of the class randomized drug-drug interaction trial. material entities, as opposed to the representation of information about physiological processes has been a core To represent the scientific observations and their strategy in developing DIDEO. properties, we imported terms from the following ontologies: Chemical Entities of Biological Importance (ChEBI) [10], In this poster we present the development of a new, Drug Ontology (DRON) [11], Gene Ontology (GO) [12], semantically rich OWL representation of types of evidence for IV. CONCLUSION Based on these results we conclude that the attributes of evidence as used by the DIKB are sufficient to infer a taxonomy of evidence types automatically. We also conclude that it is feasible to use these attributes to automatically categorize individual evidence items using OWL reasoning. ACKNOWLEDGEMENT For all authors: This project is supported by a grant from the National Library of Medicine: “Addressing gaps in clinically useful evidence on drug-drug interactions” (R01LM011838). JS is supported by training grant T15LM007059 from the National Library of Fig. 1. The formal definition of randomized drug-drug interaction trial in DIDEO. The boxes represent classes; the arrows represent object properties. Medicine/National Institute of Dental and Craniofacial All depicted object properties are used in existential statements (SOME). Research. Ontology of Adverse Events (OAE) [13], Ontology of REFERENCES Biomedical Investigations (OBI) [14], and the Uberon multi- [1] R. Boyce, C. Collins, J. Horn, I. Kalet, "Computing with evidence: Part species anatomy ontology [15]. I," Journal of Biomedical Informatics 42(6), pp. 979–989, 2009. [2] R. Boyce, C. Collins, J. Horn, I. Kalet, "Computing with evidence: Part III. RESULTS II," Journal of Biomedical Informatics 42(6), pp. 990–1003, 2009. The extension of DIDEO currently available includes 24 [3] P. Grenon, B. Smith, L. Goldberg, “Biodynamic Ontology: Applying formally defined evidence types. It can be accessed from BFO in the Biomedical Domain”, in Ontologies in Medicine: Proceedings of the Workshop on Medical Ontologies, Rome October http://purl.obolibrary.org/obo/dideo/2016-05-12/dideo.owl. 2003 (Studies in Health and Technology Informatics, 102), D. M. Representation of additional evidence types and additional Pisanelli, Ed. Amsterdam: IOS Press, 2004, pp. 20–38. axioms is underway for our project and will be implemented in [4] B. Smith, M. Ashburner, C. Rosse, J. Bard, W. Bug, W. Ceusters, et al., a subsequent version of DIDEO. “The OBO Foundry: Coordinated Evolution of Ontologies to Support Biomedical Data Integration”, Nature Biotechnology, 25 (11), pp. 1251- Running the HermiT 1.3.8.3 reasoner, we generate the 1255, November 2007. inferred hierarchy of the evidence types: it is an exact match to [5] OBO Foundry Principles. http://obofoundry.org/principles/fp-000- the previous DIKB taxonomy as built by domain experts (Fig. summary.html. Last accessed June 17, 2016 2). In addition, the example individuals were correctly sorted [6] Web Ontology Language (OWL) 2 Overview. into the evidence types based on the specified properties of the http://www.w3.org/TR/owl2-overview. Last accessed May 12, 2016 scientific observation that the evidence type was about. This [7] Brochhausen, M., Schneider, J., Malone, D., Empey, P., Hogan, W.R., result can be recreated by the reader by running the HermiT Boyce, R.D.: Towards a foundational representation of potential drug- drug interaction knowledge. in Drug Interaction Knowledge 1.3.8.3 reasoner over the test file including examples of Representation at ICBO, R. D. Boyce, M. Brochhausen, P. E. Empey, evidence items. This test file can be found here: W. R. Hogan, D. C. Malone, Eds.2014, pp. 16-31. http://ceur- http://purl.obolibrary.org/obo/dideo/EvidenceTypes/dideo.owl. ws.org/Vol-1309/paper2.pdf. Last accessed June 17, 2016 [8] B. Smith and W. Ceusters, “Ontological Realism as a Methodology for Coordinated Evolution of Scientific Ontologies”, Applied Ontology, 5, pp. 139–188, 2011. [9] A Draft Evidence Taxonomy and Inclusion Criteria for the Drug Interaction Knowledge Base (DIKB). http://purl.net/net/drug- interaction-knowledge-base/evidence-types-and-inclusion-criteria. Last accessed June 17, 2016 [10] Chemical Entities of Biological Interest (ChEBI), http://purl.obolibrary.org/obo/chebi.owl. Last accessed May 12, 2016 [11] Drug Ontology (DRON), http://purl.obolibrary.org/obo/dron.owl. Last accessed June 17, 2016 [12] Gene Ontology (GO), http://purl.obolibrary.org/obo/go.owl. Last accessed June 17, 2016 [13] Ontology of Adverse Events (OAE), http://purl.obolibrary.org/obo/oae.owl. Last accessed June 17, 2016 [14] Ontology of Biomedical Investigations, http://purl.obolibrary.org/obo/obi.owl. Last accessed June 17, 2016 [15] Uberon multi-species anatomy ontology. http://purl.obolibrary.org/obo/uberon.owl. Last accessed June 17, 2016 Fig. 2. View of the inferred evidence type taxonomy in Protégé