Inferring potential small molecule-miRNA association based on triple layer heterogeneous network.

Area of Science:

• Biomedical Informatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are crucial regulators in human complex diseases, acting as small molecule (SM) drug targets.
• Identifying SM-miRNA associations is vital for disease therapy and clinical applications of existing drugs.
• Computational models are increasingly used to predict these associations, aiding medical research.

Purpose of the Study:

• To propose a novel computational model, TLHNSMMA, for predicting small molecule-miRNA associations.
• To integrate diverse biological data, including SM and miRNA similarities, disease similarities, and known associations, into a heterogeneous graph.
• To evaluate the predictive performance of TLHNSMMA against existing methods.

Main Methods:

• Developed a triple layer heterogeneous network-based model (TLHNSMMA).
• Integrated SM similarity, miRNA similarity, disease similarity, and known SM-miRNA and miRNA-disease associations.
• Employed cross-validation techniques (global, local leave-one-out, fivefold) for performance evaluation.
• Compared TLHNSMMA with the SMiR-NBI computational model.

Main Results:

• TLHNSMMA achieved high Area Under the Curve (AUC) values, e.g., 0.9859 on Dataset 1.
• Performance metrics demonstrated TLHNSMMA's superiority over the SMiR-NBI model.
• Case studies confirmed experimentally validated SM-miRNA associations among the top predictions.

Conclusions:

• TLHNSMMA is an effective computational tool for predicting small molecule-miRNA associations.
• The model's ability to integrate heterogeneous data enhances prediction accuracy.
• TLHNSMMA holds significant potential for drug discovery and therapeutic applications in complex diseases.