Drug repurposing with graph neural networks Towards treatment for COVID-19
Keywords
Loading...
Authors
Issue Date
2022-07-17
Language
en
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Drug repurposing has great promises for the field of drug development as it reduces the cost and
duration of the development process as well as reducing potential risks of this process. The recently
released Drug Repurposing Knowledge Graph (DRKG) combines information from existing
databases, making it a generic biomedical knowledge graph for drug repurposing [1]. Recent advances
in artificial intelligence (AI) have made it easier than ever to analyse these big biomedical
knowledge graphs. This thesis investigated whether a relational graph autoencoder (R-GAE) can
be used to derive potential drug repurposing candidates for COVID-19 from a biomedical knowledge
graph. During the thesis, it was investigated whether enhancing a generic knowledge graph
with disease-specific data results in a better and more relevant prediction of potential drug repurposing
candidates. The results suggest that adding disease-specific data to a generic knowledge
graph results in different predicted drug repurposing candidates compared to a generic knowledge
graph. Even though there are pitfalls in the approach used in this study, the attempts that have
been made offer promising insights in the use of graph neural networks for drug repurposing. Future
research will have to investigate how to overcome the limitations towards better predictions
on disease-specific drug repurposing candidates.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen