Predicting miRNA-disease association from heterogeneous information network with GraRep embedding model.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are crucial in disease development and treatment.
• Experimental methods for identifying miRNA-disease associations are costly and time-consuming.
• Computational approaches are needed for efficient prediction of miRNA-disease links.

Purpose of the Study:

• To develop a novel computational method for predicting potential miRNA-disease associations.
• To integrate heterogeneous biological network information for improved prediction accuracy.
• To validate the proposed method through cross-validation and case studies.

Main Methods:

• Constructed a heterogeneous information network including lncRNA, drug, protein, disease, and miRNA.
• Applied the Learning Graph Representations with Global Structural Information (GraRep) network embedding method.
• Integrated node embeddings with miRNA sequence and disease semantic similarity for feature representation.
• Utilized a Random Forest (RF) classifier for predicting miRNA-disease associations.

Main Results:

• Achieved 85.11% prediction accuracy, 80.41% sensitivity, and 91.25% AUC via 5-fold cross-validation.
• Validated top predicted miRNAs for Colon, Breast, and Esophageal Neoplasms against existing databases (45, 42, and 44 confirmed, respectively).

Conclusions:

• The proposed network embedding-based heterogeneous information integration method is effective for predicting miRNA-disease associations.
• This computational approach offers a powerful and useful tool for advancing miRNA-disease association research.
• The findings highlight the potential of integrating diverse biological data for disease association studies.