NEMPD: a network embedding-based method for predicting miRNA-disease associations by preserving behavior and...

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are crucial non-coding RNAs involved in biological processes and human diseases.
• Accurate identification of miRNA-disease associations aids disease research and treatment.
• Existing methods often overlook interactions with other molecules.

Purpose of the Study:

• To develop a network embedding-based method for predicting potential miRNA-disease associations.
• To integrate behavior and attribute information for improved prediction accuracy.
• To address limitations of methods using only single association types.

Main Methods:

• Constructed a heterogeneous network integrating miRNA, protein, and disease associations.
• Applied the GraRep network representation method to learn behavioral information.
• Combined behavioral and attribute information to represent miRNA-disease pairs.
• Utilized the Random Forest algorithm for the prediction model.

Main Results:

• The NEMPD model achieved an average prediction accuracy of 85.41% and 80.96% sensitivity at an AUC of 91.58% via five-fold cross-validation.
• Case studies confirmed the model's efficacy, with high validation rates for top predicted miRNA-disease associations.
• Successfully identified numerous potential miRNA-disease associations, validated against existing databases.

Conclusions:

• The proposed NEMPD model demonstrates strong performance in predicting miRNA-disease associations.
• This method holds significant potential for advancing miRNA-disease association research.
• The integrated network approach offers a promising direction for future computational predictions.