Biomedical Data Translator Technical Feasibility Assessment of Reasoning Tool: University of North Carolina at Chapel Hill

Univ Of North Carolina Chapel Hill, Chapel Hill NC

Investigators

Linked publications & trials

Abstract

Our multidisciplinary team has initiated the development of a novel Translator Reasoning Tool? Reasoning Over Biomedical Objects linked in Knowledge-Oriented Pathways (ROBOKOP)?to address, in a fully automated way, user queries that can be answered by concurrent reasoning across Knowledge Sources (KSs) by means of a Knowledge Graph (KG). Our proof-of-concept (POC) product, PROTOCOP, represents a generalizable solution for reasoning across KSs to identify answers to clinical and translational research questions such as establishing data-supported Clinical Outcomes Pathways (COPs) of drugs or pathways explaining disease-disease interactions. PROTOCOP incorporates several KSs identified by the Translator project (e.g., chem2bio2rdf, Biolink, Pharos), as well as additional KSs that may be of value to the broader Translator program such as Chemotext (chemotext.mml.unc.edu), which is a publiclyavailable webserver that mines the published literature in PubMed in the form of Medical Subject Headings (MeSH) terms. PROTOCOP has successfully identified solutions to the initial two translational query types: (1) genetic conditions that protect against disease; and (2) COPs for drug-condition pairs. Over the 10-month project period, we will refine and evaluate our approach as we move from PROTOCOP to ROBOKOP and embed it in fully operational and user-friendly software. We will develop automated approaches to identify high-scoring connected subgraphs that translate into well-reasoned, data-supported, biological pathways that explain drug action and disease mechanisms and that may help elucidate new drug targets. In addition, we will expand the set of allowable queries to include those that rely on longitudinal clinical data and time- and geospatial-dependent socio-environmental exposures data. We expect that once developed, ROBOKOP will enable novel, testable hypothesis generation for a number of biomedical challenges such as drug repurposing, detection of disease comorbidities, mechanistic interpretation of disease phenotypes in terms of underlying endotypes, and others.