A network embedding-based multiple information integration method for the MiRNA-disease association prediction.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• MicroRNAs (miRNAs) are crucial regulators of biological processes.
• Predicting miRNA-disease associations aids understanding of human disease mechanisms.
• Existing methods underutilize known miRNA-disease associations.

Purpose of the Study:

• To propose a novel network embedding-based method (NEMII) for predicting miRNA-disease associations.
• To effectively integrate multiple biological information sources for improved prediction accuracy.
• To leverage network embedding techniques for enhanced analysis of miRNA-disease relationships.

Main Methods:

• Formulated known miRNA-disease associations as a bipartite network.
• Employed Structural Deep Network Embedding (SDNE) to learn node embeddings.
• Integrated miRNA-family associations and disease semantic similarities with network embeddings.
• Utilized random forest models for prediction based on integrated features.

Main Results:

• NEMII achieved high-accuracy prediction of miRNA-disease associations.
• NEMII outperformed existing state-of-the-art methods (GRNMF, NTSMDA, PBMDA).
• Case studies validated the practical utility of NEMII.
• SDNE demonstrated superior performance compared to other network embedding methods (DeepWalk, HOPE, LE).

Conclusions:

• NEMII is a promising new method for miRNA-disease association prediction.
• The study highlights the potential of network embedding in this field.
• NEMII offers a valuable tool for advancing research in miRNA-disease associations.