Identifying circRNA-disease association based on relational graph attention network and hypergraph attention network

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Circular RNAs (circRNAs) exhibit synergistic relationships with microorganisms, potentially influencing disease development.
• Experimental validation of circRNA-disease associations is resource-intensive, necessitating advanced computational approaches.

Purpose of the Study:

• To develop and validate a novel computational model, HAGACDA, for predicting associations between circRNAs and diseases.
• To leverage multi-source biological information and advanced network analysis for accurate circRNA-disease association inference.

Main Methods:

• Feature aggregation using singular value decomposition and Pearson similarity.
• Construction of a circRNA-miRNA-disease multi-source heterogeneous network.
• Application of relational graph attention networks and hypergraph attention networks for feature extraction.
• Utilizing a multilayer perceptron for final association prediction.

Main Results:

• The HAGACDA model demonstrated high accuracy in predicting circRNA-disease associations.
• Comparative experiments confirmed the model's superior performance over existing methods.
• Case studies further validated the biological relevance of the predicted associations.

Conclusions:

• HAGACDA offers an efficient and accurate computational tool for identifying circRNA-disease associations.
• The model's approach integrates diverse biological data and advanced network analysis techniques.
• This work facilitates a deeper understanding of circRNA roles in diseases and reduces experimental costs.