DEJKMDR: miRNA-disease association prediction method based on graph convolutional network.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are critical in complex human diseases.
• Accurate identification of miRNA-disease associations is vital for disease treatment.
• Traditional methods for miRNA-disease association prediction are limited by cost and sample size.

Purpose of the Study:

• To propose a computational model for accurate miRNA-disease association prediction.
• To leverage biomolecular information for enhanced prediction accuracy.
• To overcome limitations of traditional prediction methods.

Main Methods:

• Developed DEJKMDR, a graph convolutional network (GCN)-based model.
• Integrated miRNA functional similarity, disease semantic similarity, and Gaussian interaction attributes.
• Employed DropEdge for regularization and JK-Net for adaptive learning.

Main Results:

• DEJKMDR demonstrated superior accuracy and reliability compared to existing algorithms.
• Achieved an average Area Under the Curve (AUC) of 0.9772 in 10-fold cross-validation.
• Successfully predicted unknown miRNA-disease relationships.

Conclusions:

• DEJKMDR offers a robust and accurate computational approach for miRNA-disease association prediction.
• The model's integration of diverse biomolecular data enhances its predictive power.
• This method holds promise for advancing complex disease research and therapeutic development.