Discovery Informatics over the Biomedical Semantic Web: Automating Data and Knowledge Discovery for Drug Repurposing

Date: July 8, 2016

Speaker: Maulik Kamdar, BMIR, Stanford University

Abstract

Drug repurposing research relies on the ability to discover and correlate data from several heterogeneous sources simultaneously, and requires a priori knowledge of drugs, proteins, pathways and phenotypes. Biomedical researchers need increased skills and knowledge to query multiple data sources and knowledge bases with varying schema and representation formats, and to integrate and analyze the retrieved results.

To solve the challenges of integrative bioinformatics, researchers have started using Semantic Web Technologies to develop the Life Sciences Linked Open Data (LSLOD) network. However, the current state of the LSLOD network is still unusable for most biomedical researchers due to architectural issues and learning requirements. Augmenting the LSLOD network with a natural language, federated querying method will provide a usable and a sustainable framework to the biomedical researcher for precise question–answering and automated data discovery, and enable the evaluation of hypotheses on drug repurposing.

When completed, my research will provide a solution to tackle the data discovery bottleneck in the scientific innovation process, provide an interactive interface for answering biomedical queries such as "List antineoplastic agents that target IDH1 or MGMT, with literature citations and their upstream and downstream targets". Using data from the LSLOD network, I also propose to evaluate a novel method to discover hidden drug-disease associations, in the context of Glioblastoma Multiforme.