Identification of Disease-Associated MicroRNAs Via Locality-Constrained Linear Coding-Based Ensemble Learning.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• MicroRNA (miRNA) dysregulation is linked to disease development.
• Predicting miRNA-disease associations aids in understanding disease pathogenesis.
• Wet-lab methods are resource-intensive; computational models offer a cost-effective alternative.

Purpose of the Study:

• To develop a computational model for predicting miRNA-disease associations.
• To improve the accuracy and comprehensiveness of miRNA-disease association predictions.

Main Methods:

• Proposed a novel method called locality-constrained linear coding for predicting associations (ILLCEL).
• Utilized miRNA sequence similarity, miRNA functional similarity, disease semantic similarity, and interaction profile similarity as prior information.
• Employed an ensemble learning framework with hypergraph-regular terms for enhanced prediction accuracy.

Main Results:

• ILLCEL demonstrated superior prediction performance in fivefold cross-validation.
• The model accurately predicted known miRNA-disease associations.
• Novel associations were successfully verified in databases (HMDD v3.2, miRCancer) and literature.

Conclusions:

• ILLCEL is a powerful and accurate computational tool for inferring potential miRNA-disease associations.
• The method provides a cost-effective complement to traditional experimental approaches.
• This work contributes to a deeper understanding of miRNA roles in disease.