Using Graph Attention Network and Graph Convolutional Network to Explore Human CircRNA-Disease Associations Based on...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Circular RNAs (circRNAs) play a role in disease pathogenesis.
• Understanding circRNA-disease links is vital for developing new treatments.
• Existing computational models struggle with multisource data and sparse networks.

Purpose of the Study:

• To develop an advanced computational method for detecting circRNA-disease relationships.
• To effectively utilize multisource biomedical data for improved prediction accuracy.

Main Methods:

• Developed GATGCN, integrating Graph Attention Network (GAT) and Graph Convolutional Network (GCN).
• Fused multiple biomedical data sources using Centered Kernel Alignment (CKA) for data weighting.
• Employed GAT for latent representation learning and GCN for feature extraction via neighbor aggregation.

Main Results:

• GATGCN achieved high performance with an AUC of 0.951 (leave-one-out) and 0.932 (5-fold cross-validation).
• Case studies on lung cancer, diabetes retinopathy, and prostate cancer confirmed GATGCN's reliability.
• The method effectively handles sparse networks and integrates heterogeneous data.

Conclusions:

• GATGCN is a robust and effective computational tool for predicting circRNA-disease associations.
• The approach offers a significant advancement in leveraging complex biomedical data for disease mechanism exploration.
• This method holds promise for identifying novel therapeutic targets and understanding disease pathways.