Dual-Network Collaborative Matrix Factorization for predicting small molecule-miRNA associations

Area of Science:

• Biomedical informatics
• Computational biology
• Pharmacogenomics

Background:

• MicroRNAs (miRNAs) are key regulators in biological processes and diseases.
• Small molecules (SMs) targeting miRNAs offer therapeutic potential.
• Experimental validation of SM-miRNA associations is costly and time-consuming.

Purpose of the Study:

• To develop a novel computational method for predicting potential small molecule-microRNA associations.
• To address the limitations of experimental validation through an efficient predictive model.

Main Methods:

• Proposed Dual-network Collaborative Matrix Factorization (DCMF) model.
• Utilized Weighted K Nearest Known Neighbors (WKNKN) for data preprocessing.
• Employed matrix factorization to derive latent features of SMs and miRNAs.
• Integrated dual network information for enhanced similarity analysis.

Main Results:

• DCMF achieved high prediction accuracy across four cross-validation methods on two datasets.
• Area Under the Curves (AUC) ranged from 0.8377 to 0.9868.
• Case studies confirmed numerous predicted SM-miRNA associations with published literature.

Conclusions:

• DCMF is an effective computational tool for predicting small molecule-microRNA associations.
• The method offers a cost-efficient and rapid approach compared to experimental validation.
• DCMF facilitates the identification of novel therapeutic strategies targeting miRNAs.