Prediction of circRNA-Disease Associations via Graph Isomorphism Transformer and Dual-Stream Neural Predictor

Area of Science:

• Biochemistry
• Genomics
• Computational Biology

Background:

• Circular RNAs (circRNAs) play significant roles in human diseases.
• Predicting circRNA-disease associations (CDAs) is vital for disease diagnosis and treatment.
• Existing computational methods face challenges with data sparsity and effective prediction.

Purpose of the Study:

• To develop a novel computational method for predicting circRNA-disease associations (CDAs).
• To address data sparsity and improve the accuracy of CDA prediction.
• To leverage knowledge graph technology for enhanced CDA discovery.

Main Methods:

• Constructed a multi-source heterogeneous knowledge graph incorporating circRNAs, diseases, and other non-coding RNAs (lncRNAs, miRNAs).
• Employed a Graph Isomorphism Transformer (GIT) to analyze local and global association information within the knowledge graph.
• Utilized a Dual-Stream Neural Predictor (DSP) to integrate dual-stream features for accurate CDA prediction.

Main Results:

• The proposed Graph Isomorphism Transformer with Dual-Stream Neural Predictor (GIT-DSP) method demonstrated superior performance compared to existing state-of-the-art models.
• GIT-DSP effectively addressed data sparsity issues inherent in CDA prediction.
• The model successfully uncovered potential circRNA-disease associations, offering deeper insights.

Conclusions:

• GIT-DSP provides a powerful and accurate approach for predicting circRNA-disease associations.
• The method enhances our understanding of the roles circRNAs play in diseases.
• Findings contribute valuable insights for precision medicine and disease-related research.