Predicting lncRNA-disease associations using network topological similarity based on deep mining heterogeneous...

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• Long noncoding RNAs (lncRNAs) are crucial in biological processes and disease, yet their associations with diseases are not fully understood.
• Predicting lncRNA-disease associations is vital for understanding disease mechanisms and developing new biomarkers.
• Current methods are limited, necessitating novel approaches for accurate prediction.

Purpose of the Study:

• To develop a novel computational method for predicting lncRNA-disease associations.
• To leverage heterogeneous biological network data and the DeepWalk algorithm for improved prediction accuracy.
• To identify potential novel associations between lncRNAs and diseases.

Main Methods:

• Constructed a Linked Tripartite Network integrating known lncRNA-disease, lncRNA-microRNA, and microRNA-disease interactions.
• Applied the DeepWalk algorithm to extract topological structure features from the network nodes.
• Calculated lncRNA-lncRNA and disease-disease similarities based on network topology.
• Utilized a rule-based inference method to discover new lncRNA-disease associations.

Main Results:

• The proposed method demonstrated superior predictive performance in identifying lncRNA-disease associations.
• The approach effectively utilizes network topology for similarity measurement, offering a novel perspective beyond sequence or structure.
• The Area Under the Curve (AUC) value confirmed the method's predictive power.

Conclusions:

• The DeepWalk-based method provides a powerful and novel approach for predicting lncRNA-disease associations.
• Integrating heterogeneous biological data into a network framework enhances the discovery of complex molecular interactions.
• This method offers a valuable tool for prioritizing potential lncRNA-disease links for experimental validation.