Predicting associations between CircRNA and diseases through structure-aware graph transformer and path-integral...

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Circular RNAs (circRNAs) are critical regulators in cellular processes and implicated in various diseases.
• Traditional experimental methods for identifying circRNA-disease associations are inefficient and costly.
• Existing computational methods lack the ability to capture structural similarities between biological network edges.

Purpose of the Study:

• To develop an advanced computational method, SATPIC2CD, for analyzing potential associations between circular RNAs and diseases.
• To overcome the limitations of existing nodal-perspective computational approaches.
• To improve the accuracy and efficiency of circRNA-disease association prediction.

Main Methods:

• Utilized a Structure-Aware Graph Transformer (SAT) to extract metapath representations and integrate structural information.
• Employed Path Integral Convolutional Networks (PACN) for integrating feature information across path weights.
• Incorporated Gated Recurrent Units (GRU) and node centrality for feature loss and smoothing, followed by a Multi-Layer Perceptron (MLP) for prediction.

Main Results:

• SATPIC2CD achieved a high accuracy of 0.9715 AUC in 5-fold cross-validation.
• The method outperformed existing comparative models in predicting circRNA-disease associations.
• Case studies demonstrated the high precision of SATPIC2CD in identifying these associations.

Conclusions:

• SATPIC2CD is a highly accurate and efficient computational tool for predicting circular RNA-disease associations.
• The method provides a strong foundation for advancing biological research and understanding disease mechanisms.
• This approach enhances the identification of potential therapeutic targets and strategies.