DETHACDA: A Dual-View Edge and Topology Hybrid Attention Model for CircRNA-Disease Associations Prediction

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Circular RNAs (circRNAs) play roles in complex diseases and are potential therapeutic targets.
• Experimental identification of disease-associated circRNAs is slow and resource-intensive.
• Existing computational models for circRNA-disease association often overlook crucial semantic information.

Purpose of the Study:

• To develop an intelligent and precise computational model for predicting circRNA-disease associations.
• To improve upon existing graph-based methods by incorporating both network topology and semantic features.
• To identify novel circRNA-disease associations for therapeutic development.

Main Methods:

• Proposed a Dual-view Edge and Topology Hybrid Attention model (DETHACDA).
• Utilized a heterogeneous network to capture neighborhood topology and diverse semantics of circRNA and disease nodes.
• Employed 5-fold cross-validation for performance evaluation.

Main Results:

• DETHACDA achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.9882 on the circRNADisease dataset.
• The model significantly outperformed four state-of-the-art computational methods.
• Demonstrated the effectiveness of integrating topological and semantic information for accurate prediction.

Conclusions:

• DETHACDA provides a highly accurate and efficient method for predicting circRNA-disease associations.
• The model's ability to capture complex network features enhances its predictive power.
• This approach facilitates the identification of circRNAs as potential biomarkers and therapeutic targets for complex diseases.