Program 
Abstract
Creation and exploration of a semantic knowledge base for mechanism-based Inference between diseases, gene networks and drugs
 
Bruce J Aronow, PhD
Divisions of Biomedical Informatics and Developmental Biology Scientific Director, Center for Computational Medicine Cincinnati Children's Hospital Medical Center University of Cincinnati College of Medicine
 
Finding connections between small molecules such as existing drug and poorly treated diseases is an attractive strategy for translational research. However, the identification of such connections remains highly dependent on serendipitous observations and educated guesses. To pursue a systematic approach to the discovery of novel and inferable relationships between drugs and diseases based on mechanistic knowledge, we have sought to develop a semantic infrastructure that could allow integration of heterogeneous data from pharmacological and biological knowledge sets and enable efficient mining of non-trivial connections and causal relations.
To approach this we have devised a Disease-Drug Correlation Ontology (DDCO), an ontological framework to allow integration of varied datasets extracted from pharmacological and biological domains. The DDCO, formalized in OWL, is a result of integrating multiple sources of ontologies, controlled vocabularies, and data schemas. We demonstrate how the resultant DDCO framework supports integration and representation of a collection of data sources including DrugBank, EntrezGene, OMIM, KEGG, BioCarta, Reactome, and UMLS, and construction of a comprehensive pharmacome-genome-diseasome network into RDF. Using a disease centered approach from knowledge associate with the disease Systemic Lupus Erythematosus (SLE), we performed ontology-driven data query to form a moderately large-scale RDF network, and then applied several graph structure analytic approaches to derive known drugs with respect to likelihood for altering disease severity. Using this approach ranks Tamoxifen or mechanistically related compounds particularly highly, and thus provides intriguing support for several proposed early stage investigations considering the uses of Tamoxifen in SLE. The results demonstrate an exciting direction using Semantic Web methodologies for translational clinical research at the level of data integration, knowledge representation, and the enablement of graph query languages and network analysis algorithms to allow mining of drug action and disease mechanism relationships.  Our work indicates that improvements in semantic representation and data knowledge annotation of biological, therapeutic, and disease-based mechanistic relationships will provide a fertile basis for accelerated drug repositioning and discovery applications.
Biosketch
Bruce J. Aronow, PhD
Professor and Co-director, Computational Medicine Center, Cincinnati Children's Hospital Medical Center.  Bruce's research lab is devoted to unraveling both the role and mechanism by which the functional capabilities of the human genome shape human health and our ability to adapt to stressful challenges. His lab is using a wide variety of available structural and functional genomic and biological systems descriptive data to form models of how biological systems assemble, adapt and become impaired in disease. The lab's overall hypothesis is that by interconnecting as much experimental and observational information as possible, they can gain new insights into the mechanisms by which different biological systems can achieve health or healthy adaptation, or undergo disease processes. Bruce earned his BS in Chemistry at Stanford University in 1976 and his PhD in Biochemistry at the University of Kentucky in 1986. He completed his Research Fellowship at the Division of Basic Science Research, Cincinnati Children's Research Foundation from 1986 to 1989.rticle