Cluster correlation based method for lncRNA-disease association prediction.

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Disease Association Studies

Background:

• Long non-coding RNAs (lncRNAs) play critical roles in human biological pathways.
• Dysregulation of lncRNAs is linked to various complex human diseases.
• Predicting lncRNA-disease associations is crucial for disease diagnosis and treatment, yet remains challenging due to unclear lncRNA functions.

Purpose of the Study:

• To develop a novel computational method for predicting potential associations between lncRNAs and diseases.
• To leverage known lncRNA-disease and gene-disease relationships for accurate prediction.
• To provide a tool for identifying novel therapeutic targets and understanding disease mechanisms.

Main Methods:

• Construction of a bipartite network integrating known disease-lncRNA and disease-protein coding gene associations.
• Calculation of cluster association scores to quantify relationships between disease and gene clusters.
• Definition of gene-disease association scores based on cluster scores to predict novel associations.

Main Results:

• The proposed method demonstrated reliable performance in Leave-One-Out Cross Validation (LOOCV) with AUCs of 0.8169 and 0.8410.
• 5-fold cross-validation yielded AUCs of 0.7573 and 0.8198, outperforming three other comparative methods.
• The method's efficiency was highlighted, with successful verification of results for melanoma and ovarian cancer in case studies.

Conclusions:

• The developed method effectively predicts lncRNA-disease associations with high accuracy and reliability.
• The approach is simple, efficient, and adaptable for incorporating new gene-disease associations.
• The findings offer valuable insights for further research into lncRNA functions and their roles in complex diseases.