KGETCDA: an efficient representation learning framework based on knowledge graph encoder from transformer for...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Circular RNAs (circRNAs) are increasingly recognized for their roles in human disease progression.
• Efficient identification of circRNA-disease associations (CDAs) is crucial for disease diagnosis and understanding.
• Existing computational methods for CDA prediction face challenges with data sparsity and capturing complex, high-order interactions.

Purpose of the Study:

• To develop a novel computational method, KGETCDA, for accurate and efficient prediction of circRNA-disease associations.
• To address limitations of existing methods, particularly in handling data sparsity and exploring high-order biological information.
• To provide a user-friendly platform (HNRBase) for accessing and utilizing the developed prediction tool and associated data.

Main Methods:

• Constructed a large heterogeneous non-coding RNA dataset by integrating over 10 databases.
• Built a biological knowledge graph incorporating relationships between circRNA, miRNA, lncRNA, and disease.
• Employed Transformer-based knowledge representation learning and attentive propagation for high-quality embedding generation.
• Utilized multilayer perceptron for predicting CDA matching scores based on learned embeddings.

Main Results:

• KGETCDA demonstrated significantly superior performance compared to state-of-the-art models in predicting circRNA-disease associations.
• The method effectively captured high-order interaction information, overcoming data sparsity limitations.
• An interactive web platform, HNRBase, was developed for data visualization, download, and prediction, enhancing usability.

Conclusions:

• KGETCDA offers a powerful and accurate computational approach for identifying circRNA-disease associations.
• The integration of biological knowledge graphs and advanced learning techniques enhances prediction capabilities.
• HNRBase provides a valuable resource for researchers studying circRNA functions in human diseases.