Predicting CircRNA-Disease Associations Based on Heterogeneous Graph Neural Network and Knowledge Graph Attribute...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Understanding circular RNA (circRNA) and disease associations is crucial for disease pathogenesis research.
• Existing computational methods for identifying circRNA-disease associations have limitations, particularly in handling noise.
• There is a need for robust methods to accurately predict circRNA-disease associations.

Purpose of the Study:

• To propose a novel computational method, KAATCDA, for predicting circRNA-disease associations.
• To address the limitations of existing methods by incorporating knowledge graph attributes and attention mechanisms.
• To improve the accuracy and reliability of circRNA-disease association identification.

Main Methods:

• Developed the Knowledge Graph Attribute Mining Attention Network (KAATCDA) for circRNA-disease association prediction.
• Utilized a Knowledge Graph Attribute network (KGA) to learn disease feature representations.
• Employed an Attribute Mining Attention network (AMA) to obtain circRNA features aligned with disease representations.
• Predicted circRNA-disease association scores based on learned feature representations.

Main Results:

• KAATCDA demonstrated superior performance compared to state-of-the-art methods in five-fold cross-validation experiments on two datasets.
• Experimental results indicate that the proposed method effectively identifies circRNA-disease associations.
• A case study confirmed the method's capability in predicting previously unknown circRNA-disease associations.

Conclusions:

• KAATCDA offers an effective and accurate approach for predicting circRNA-disease associations.
• The method's ability to handle noise and learn robust feature representations contributes to its improved performance.
• This work advances the understanding of circRNA's role in disease pathogenesis and provides a valuable tool for researchers.