MCCMF: collaborative matrix factorization based on matrix completion for predicting miRNA-disease associations.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are non-coding RNAs regulating gene expression.
• miRNA-disease associations are crucial for understanding human diseases.
• Traditional methods for identifying these associations are slow and imprecise.

Purpose of the Study:

• To propose a novel computational method, MCCMF, for predicting unknown miRNA-disease associations.
• To overcome the limitations of existing models in terms of speed and accuracy.

Main Methods:

• Utilized matrix completion to reconstruct the miRNA-disease association matrix.
• Incorporated Gaussian Interaction Profile kernel for miRNA functional and disease semantic similarity.
• Applied Weight K Nearest Known Neighbors for data preprocessing.
• Employed collaborative matrix factorization for prediction.

Main Results:

• Achieved a satisfactory Area Under the Curve (AUC) of 0.9569 (±0.0005) in fivefold cross-validation.
• Demonstrated superior performance compared to other advanced methods.
• Evaluated using accuracy, precision, recall, and f-measure.

Conclusions:

• MCCMF significantly outperforms existing methods in predicting miRNA-disease associations.
• The model's effectiveness and practicality are validated through case studies on three specific diseases.
• Provides a more accurate and efficient tool for miRNA-disease association research.