PDDGCN: A Parasitic Disease-Drug Association Predictor Based on Multi-view Fusion Graph Convolutional Network

Area of Science:

• Computational biology
• Parasitology
• Drug discovery

Background:

• Accurate identification of disease-drug associations is crucial for understanding parasitic diseases.
• Existing computational methods often rely on limited link-based approaches in bipartite networks.

Purpose of the Study:

• To develop an advanced computational model for predicting associations between parasitic diseases and drugs.
• To improve upon existing methodologies for disease-drug association prediction.

Main Methods:

• Proposed PDDGCN, a multi-view graph convolutional network model.
• Fused similarity and binary networks into multi-view heterogeneous networks.
• Employed neighborhood information aggregation and inter/intra-domain message passing for node embedding refinement.

Main Results:

• PDDGCN demonstrated superior performance compared to five state-of-the-art methods and four machine learning algorithms.
• Experimental results validated the model's effectiveness in identifying parasitic disease-drug associations.
• Case studies confirmed the practical utility of PDDGCN.

Conclusions:

• The PDDGCN model offers a promising approach for discovering novel treatments for parasitic diseases.
• This work contributes to a deeper understanding of the etiology of parasitic diseases.
• The developed model can accelerate drug discovery and therapeutic development in parasitology.