An improved random forest-based computational model for predicting novel miRNA-disease associations

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) regulate gene expression post-transcriptionally; their dysregulation is linked to complex human diseases.
• Accurate identification of disease-associated miRNAs is crucial for understanding disease mechanisms and developing treatments.
• Computational models offer an economical and effective alternative to time-consuming and expensive experimental methods for predicting miRNA-disease associations.

Purpose of the Study:

• To develop an improved computational model, IRFMDA (Identifying miRNA-Disease Associations), for predicting associations between miRNAs and diseases.
• To enhance the accuracy of miRNA-disease association prediction by integrating multiple similarity measures and employing feature selection.

Main Methods:

• Integrated disease similarity was calculated using semantic and Gaussian interaction profile kernel (GIPK) similarity.
• Integrated miRNA similarity was computed using functional and GIPK similarity.
• A random forest (RF) model was trained using integrated similarities, incorporating feature selection based on variable importance scores for optimizing prediction.

Main Results:

• The IRFMDA model achieved high performance with AUCs of 0.8728 (local LOOCV), 0.9398 (global LOOCV), and 0.9363 (5-fold CV), outperforming existing models.
• Case studies on esophageal cancer, lymphoma, and lung cancer showed high validation rates for predicted miRNA-disease associations against experimental data.
• IRFMDA successfully identified 94, 98, and 100 top-ranked disease-associated miRNAs for esophageal cancer, lymphoma, and lung cancer, respectively, supported by dbDEMC v2.0.

Conclusions:

• IRFMDA demonstrates excellent performance as a miRNA-disease association prediction model.
• The model's accuracy and validation suggest its utility in guiding future experimental research.
• IRFMDA can provide valuable insights into the regulatory roles of miRNAs in complex human diseases.