MGCNRF: Prediction of Disease-Related miRNAs Based on Multiple Graph Convolutional Networks and Random Forest

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are increasingly recognized for their role in various diseases.
• Traditional experimental methods for identifying miRNA-disease associations (MDAs) are time-consuming and laborious.
• Existing computational methods for MDA prediction have limitations in performance and accuracy.

Purpose of the Study:

• To develop an advanced computational model for predicting miRNA-disease associations (MDAs).
• To overcome the limitations of current methods in terms of predictive performance and accuracy.
• To provide a reliable computational tool for identifying potential MDAs.

Main Methods:

• A novel model, multiple graph convolutional networks and random forest (MGCNRF), was developed.
• MGCNRF constructs four two-layer heterogeneous networks integrating miRNA and disease data.
• It utilizes layered attention graph convolutional networks (GCNs) to extract embeddings and predicts MDAs using random forest (RF).

Main Results:

• MGCNRF demonstrated superior prediction performance compared to seven state-of-the-art methods in fivefold cross-validation.
• The model achieved a high area under the curve (AUC), indicating strong predictive power.
• Case studies validated the scientific rationale and accuracy of MGCNRF for predicting potential MDAs.

Conclusions:

• MGCNRF offers a significant advancement in computational prediction of miRNA-disease associations (MDAs).
• The model's high accuracy and validated performance establish it as a valuable scientific tool.
• MGCNRF can accelerate the discovery of novel MDAs, aiding disease research and treatment strategies.