Abstract

Millions of Americans suffer from illnesses with non-existent or ineffective drug treatment. Identifying plausible drug candidates is a major barrier to drug development due to the large amount of time and resources required; approval can take years when people are suffering now. Though computational tools can expedite drug candidate discovery, utilizing these tools typically requires programming expertise that many biologists lack. This problem persists due to an ever increasing growth of new biomedical data that is difficult to integrate and maintain; such tools very seldom provide a non-programming interface for researchers to query. This creates an opportunity for a suite of user-friendly software tools to aid computational discovery of new treatments using existing drugs, eliminating the need for researchers to acquire computational expertise in integrating multiple databases and performing algorithmic analysis. Our team has unique biomedical knowledge and software development expertise through our affiliation with the Stanford School of Medicine and Department of Computer Science. Specifically, our aims are to:

• build a computational knowledge graph focused on drugs, genes, proteins, and diseases (the components relevant to identifying drug targets) with information from existing biomedical databases
• build a web interface on top of our knowledge graph that lists the connections between a specific drug, gene, or protein to all related drugs, genes, proteins, and diseases in the network (e.g., the tool would list drugs relevant to breast cancer when given the source disease ‘breast cancer’ and target node type ‘drug’)
• build a web interface on top of our knowledge graph that when given a specific drug, gene, or protein, identifies similar drugs, genes, proteins, and diseases to suggest candidates for drug targets
• evaluate utility to biomedical researchers by asking them to use the two tools to query drug discovery related drugs, genes, proteins, and diseases, and surveying them on content usefulness and tool ease of use.

The completion of this project will allow researchers direct access to comprehensive biomedical data through intuitive software that aids in decision making with regards to identifying drug candidates to provide faster, more efficacious treatment to all Americans.