Drug Repurposing Impact Project

Supported by Arnold Ventures 

Introduction

Drug repurposing offers a unique opportunity to accelerate the development of treatments for diseases with unmet medical needs. Despite its promise, one challenge in selecting repurposing opportunities lies in systematically identifying and prioritizing the most impactful ones among the vast number of potential drug-disease pairs. To address this challenge, we developed a novel scoring system to systematically identify promising drug repurposing opportunities that maximize impact on patients and economic impact on the health care system. By integrating large language models (LLMs), rule-based artificial intelligence (AI), and informatics, this system enables rapid, data-driven evaluation and scoring of opportunities. This approach facilitates the identification of high-impact candidates for further investigation, ensuring that Every Cure and other research groups can efficiently allocate resources towards the most promising therapeutic interventions. We are pleased with the support from Arnold Venture and how the support has enabled us to build a resource to be used to encourage impact-driven drug repurposing and computational pharmacophenomics.

Accomplishments (See links below for source code and download of Drug Repurposing Impact Scores)

Impact-driven drug repurposing is a new approach that we’re pioneering where you don’t begin by searching for potential repurposed treatments for a specific disease or for potential diseases that could be treated with an existing drug. Instead, you use the likelihood and amount of impact that can be made across all drug-disease matches to guide which drug-disease match to pursue. To do this work, you must have a rapid and scalable way of assessing the impact for drug repurposing opportunities as was created in the project.

In addition to developing a scoring system related to the impact of drug repurposing opportunities we have extended our work to develop a companion web application to serve as an interactive dashboard for browsing, manipulating, and downloading the data generated by the scoring algorithm. For more information, see Appendix 3 below. We created the dashboard due to an important strategic shift, that a fixed prioritization algorithm where drug/disease pairs are explicitly retained or deleted fails to account for varying priorities of the person exploring prioritization opportunities. We instead generate separate “component scores” describing different dimensions of repurposing suitability (e.g., unmet medical need, economic feasibility, regulatory feasibility, etc.) which are combined as a weighted average into an overall “priority score”. To suit different needs, the dashboard allows the user to re-weight the contribution of each component score to the final priority score. For example, there may be situations where economic feasibility is relatively unimportant to the user while unmet medical need is significantly more important.

Appendix 1 – Source code repository

All code used to create the novel drug repurposing component scores is freely and publicly available online here. 

Appendix 2 – Drug repurposing score data files

The set of all currently available repurposing prioritization component scores (minus the Biological Plausibility and Unmet Medical Need scores, which are temporarily under embargo) is available for public download. 

Appendix 3 – Interactive web application

An interactive, browser-based dashboard for rapidly browsing, visualizing, and downloading the data is publicly available, with the corresponding source code available on GitHub here.