Identification of MiRNA-Disease Associations Based on Information of Multi-Module and Meta-Path.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) regulate key biological processes like cell proliferation and apoptosis.
• Understanding miRNA-disease associations is crucial for identifying diagnostic biomarkers and therapeutic targets.
• Traditional experimental methods for identifying these associations are time-consuming and labor-intensive.

Purpose of the Study:

• To develop an efficient computational method for identifying potential associations between microRNAs and human diseases.
• To overcome the limitations of traditional experimental approaches.

Main Methods:

• Constructed a multi-module heterogeneous network using meta-paths.
• Employed Graph Attention Networks (GAT) to learn latent features of network modules.
• Averaged weighted latent features to obtain final node representations.
• Utilized Support Vector Machines (SVM) to recognize potential miRNA-disease associations based on node representations.

Main Results:

• The proposed computational method achieved high performance.
• Achieved an Area Under the Precision-Recall Curve (AUPR) of 0.9379.
• Achieved an Area Under the Receiver-Operating Characteristic Curve (AUC) of 0.9472 in five-fold cross-validation and benchmark datasets.

Conclusions:

• The developed method demonstrates outstanding practical application performance.
• Provides a valuable reference for discovering novel miRNA biomarkers and therapeutic targets.
• Highlights the potential of GAT and SVM in advancing miRNA-disease association research.