February 14th, 2020

Topic: ROBOKOP and the Biomedical Data Translator: Efficiently leveraging a distributed data ecosystem

Presenter: Dr.Karamarie Fecho - Copperline Professional Solutions; Renaissance Computing Institute, UNC Chapel Hill and Dr.Patrick Wang - CoVar Applied Technologies, Inc.

SLIDES    RECORDING

Abstract:

The Biomedical Data Translator Consortium seeks “to develop a platform that supports the translation of data into knowledge by applying inferential methods to a graph representation of biomedical entities and their relationships.” Historically, this means accessing diverse sources of biomedical data (Pharos, CTD, UniProt, …), integrating them in a monolithic knowledge graph (KG), and querying this KG to extract inferences from the data. We developed Reasoning Over Biomedical Objects linked in Knowledge Oriented Pathways (ROBOKOP) to perform these tasks, and provide a (non-technical) user interface to biomedical question-answering capabilities. Efficiently aggregating, storing, analyzing, and presenting such massive data presents software engineering challenges that we approached via containerization and parallelization. We will review the architecture we developed, highlighting the big issues encountered and/or solved, as well as ongoing development efforts.