DeepMNE: Deep Multi-Network Embedding for lncRNA-Disease Association Prediction.

Area of Science:

• Genomics
• Computational Biology
• Biomedical Informatics

Background:

• Long non-coding RNAs (lncRNAs) are implicated in human disease pathogenesis.
• Accurate identification of lncRNA-disease associations is vital for clinical applications.
• Current computational methods struggle with integrating multi-omics data and predicting novel associations.

Purpose of the Study:

• To develop a deep learning-based method, DeepMNE, for predicting lncRNA-disease associations.
• To address the challenge of identifying associations for novel lncRNAs and diseases.
• To improve the integration of multi-omics data for enhanced prediction accuracy.

Main Methods:

• DeepMNE utilizes multi-omics data to characterize diseases and lncRNAs.
• A deep learning-based network fusion approach integrates diverse data sources.
• Kernel neighborhood similarity constructs disease and lncRNA similarity networks.
• Graph embedding is employed for predicting potential lncRNA-disease associations.

Main Results:

• DeepMNE demonstrates superior predictive performance compared to existing methods, especially for novel associations, lncRNAs, and diseases.
• The method shows robust performance on perturbed datasets.
• Case studies confirm DeepMNE's efficacy as a tool for disease-related lncRNA prediction.

Conclusions:

• DeepMNE offers an effective computational strategy for discovering lncRNA-disease associations.
• The method advances the prediction of associations involving novel biological entities.
• DeepMNE provides a valuable tool for advancing research in lncRNA-mediated diseases.