SPCMLMI: A structural perturbation-based matrix completion method to predict lncRNA-miRNA interactions

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long non-coding RNA (lncRNA) and microRNA (miRNA) interactions are vital for gene regulation and complex disease development.
• Experimental identification of lncRNA-miRNA interactions is costly and time-consuming.
• Accurate computational methods are needed to infer these interactions efficiently.

Purpose of the Study:

• To develop a novel computational approach, SPCMLMI, for predicting lncRNA-miRNA interactions.
• To improve the accuracy and efficiency of identifying potential lncRNA-miRNA interactions.

Main Methods:

• Calculated lncRNA and miRNA similarities (expression profiles, sequences).
• Constructed a bilayer network integrating known interactions and similarity networks.
• Applied a structural perturbation-based matrix completion method for prediction.

Main Results:

• SPCMLMI achieved high prediction performance with AUCs of 0.8984 and 0.9891 on two datasets.
• Outperformed existing methods in prediction accuracy.
• Case studies on lncRNA XIST and miRNA hsa-mir-195-5-p validated the method's effectiveness.

Conclusions:

• SPCMLMI is an effective computational tool for predicting lncRNA-miRNA interactions.
• The method offers a valuable resource for understanding gene regulation in complex diseases.
• The structural consistency of the bilayer network contributes to prediction accuracy.