Integrating Comorbidity Knowledge for Alzheimer's Disease Drug Repurposing using Multi-task Graph Neural Network

Area of Science:

• Biomedical Informatics
• Computational Neuroscience
• Pharmacology

Background:

• Alzheimer's Disease (AD) is complex, sharing origins with comorbidities like vascular diseases.
• Leveraging knowledge from these comorbidities can aid in discovering new AD treatments.
• Existing therapeutic strategies for AD often fall short, necessitating novel approaches.

Purpose of the Study:

• To develop a computational pipeline for predicting repurposable drugs for Alzheimer's Disease (AD).
• To utilize knowledge from AD's common comorbidities, particularly vascular diseases, for drug discovery.
• To integrate genetic markers, therapeutics, and disease interactions within a graph neural network framework.

Main Methods:

• A multi-task graph neural network (GNN) pipeline was developed.
• The pipeline incorporated biomedical interactomes, genetic markers, and therapeutics of AD and related vascular diseases.
• Drug candidates were predicted using node embedding similarity within the network.

Main Results:

• The GNN pipeline accurately captured disease interactions and enabled classification.
• Predicted drug candidates demonstrated high blood-brain barrier (BBB) permeability.
• Literature review supported the potential of candidate drugs in treating AD pathologies, symptoms, or co-occurring conditions.

Conclusions:

• The proposed pipeline offers a viable strategy for repurposing drugs for AD.
• This approach effectively predicts drug candidates by analyzing biological interplays between AD and vascular diseases.
• The findings highlight the potential of computational methods in accelerating AD drug discovery.