NARRMDA: negative-aware and rating-based recommendation algorithm for miRNA-disease association prediction

Area of Science:

• Biomedical Informatics
• Genomics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are crucial in biological processes and human diseases.
• Predicting miRNA-disease associations is vital for disease research.
• Existing computational methods for miRNA-disease prediction lack reliability.

Purpose of the Study:

• To develop a novel computational model, NARRMDA, for predicting potential miRNA-disease associations.
• To enhance the accuracy and reliability of miRNA-disease association predictions.
• To provide a robust tool for identifying disease-related miRNAs.

Main Methods:

• Developed the Negative-Aware and rating-based Recommendation algorithm for miRNA-Disease Association prediction (NARRMDA).
• Utilized known miRNA-disease associations from the HMDD database.
• Incorporated miRNA functional similarity, disease semantic similarity, and Gaussian interaction profile kernel similarity.
• Employed leave-one-out cross-validation for performance evaluation.

Main Results:

• NARRMDA achieved a superior Area Under the Curve (AUC) of 0.8053, outperforming four classical prediction models.
• Case studies on colon neoplasms, esophageal neoplasms, lymphoma, and breast neoplasms showed high validation rates (92%, 84%, 92%, 100% for top 50 predictions).
• Experimental validation confirmed the reliable prediction ability of NARRMDA.

Conclusions:

• NARRMDA demonstrates superior prediction accuracy and reliability for identifying disease-associated miRNAs.
• The model offers a valuable computational approach for advancing miRNA-disease association research.
• NARRMDA can significantly aid in the discovery of novel biomarkers and therapeutic targets.