EPMDA: an expression-profile based computational model for microRNA-disease association prediction

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are crucial for biological processes and disease mechanisms.
• Existing computational models for miRNA-disease associations often rely on functional similarity and biological networks, leading to errors and bias.
• Limitations in information resources used by current methods necessitate novel approaches.

Purpose of the Study:

• To develop the first computational method, EPMDA, utilizing miRNA expression profiles for predicting miRNA-disease associations.
• To overcome the limitations of existing models by incorporating expression data.

Main Methods:

• Developed the EPMDA model, a computational approach leveraging miRNA expression profiles.
• Constructed a dataset from the HMDD v2.0 database for model training and validation.
• Performed leave-one-out and 5-fold cross-validation to assess predictive performance.

Main Results:

• EPMDA achieved high AUC values of 0.8945 (leave-one-out) and 0.8917 (5-fold cross-validation).
• Applied to Colon Neoplasms and Kidney Neoplasms, EPMDA identified top-25 miRNA lists with 80% and 88% confirmation rates, respectively.
• Comparative analysis showed EPMDA outperforms existing prediction models and classical algorithms.

Conclusions:

• EPMDA is a reliable and effective computational tool for predicting potential miRNA-disease associations.
• The model's performance validates the utility of miRNA expression profiles in this context.
• EPMDA is anticipated to significantly aid future biomedical research and discovery.