Predicting miRNA-disease associations using an ensemble learning framework with resampling method.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are critical in complex disease pathogenesis.
• Identifying miRNA-disease associations is vital for disease etiology, diagnosis, and treatment.
• Experimental methods for miRNA-disease association discovery are resource-intensive.

Purpose of the Study:

• To develop an effective computational method for predicting miRNA-disease associations.
• To address the challenge of sample imbalance in miRNA-disease association datasets.
• To identify novel disease-related miRNAs for further research.

Main Methods:

• An Ensemble learning framework with Resampling method for MiRNA-Disease Association (ERMDA) prediction was developed.
• A resampling strategy was employed to create balanced training subsets.
• Feature representations were extracted by integrating miRNA-miRNA similarities, disease-disease similarities, and known miRNA-disease associations.
• Feature selection was applied to reduce redundancy and enhance subset diversity.
• An ensemble approach with soft voting was used for final prediction.

Main Results:

• ERMDA demonstrated superior performance compared to state-of-the-art methods on both balanced and unbalanced datasets.
• Case studies on three human diseases validated ERMDA's capability in identifying potential disease-related miRNAs.
• The method proved effective in exploring the regulatory roles of miRNAs in complex diseases.

Conclusions:

• ERMDA is an effective and reliable computational tool for predicting miRNA-disease associations.
• The developed framework aids researchers in uncovering the significance of miRNAs in complex diseases.
• This approach facilitates the exploration of miRNA functions in disease development and progression.