PCDA-HNMP: Predicting circRNA-disease association using heterogeneous network and meta-path

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Circular RNAs (circRNAs) are increasingly recognized for their regulatory roles in human diseases via microRNA (miRNA) interactions.
• circRNAs show promise as disease biomarkers and therapeutic targets, but their associations with diseases (CDAs) are challenging to determine experimentally.
• Computational methods offer an efficient alternative for predicting CDAs, crucial for understanding complex diseases and advancing targeted therapies.

Purpose of the Study:

• To develop a novel computational method, PCDA-HNMP, for predicting circRNA-disease associations (CDAs).
• To leverage heterogeneous network analysis and meta-path mining to extract informative features for circRNAs and diseases.
• To improve prediction performance by incorporating miRNA-disease associations (mDAs).

Main Methods:

• Constructed a heterogeneous network integrating circRNAs, mRNAs, miRNAs, and diseases.
• Extracted meta-paths from the heterogeneous network to mine hidden associations and create meta-path-induced networks.
• Derived features from these networks, combined them with mDAs, and utilized XGBoost for CDA prediction.

Main Results:

• The PCDA-HNMP method achieved a high Area Under the Curve (AUC) of 0.9846 in five-fold cross-validation.
• The model's performance was significantly enhanced by the inclusion of miRNA-disease associations (mDAs).
• Analysis indicated that meta-paths derived from validated CDAs contributed most significantly to the prediction accuracy.

Conclusions:

• PCDA-HNMP is a highly effective computational method for predicting circRNA-disease associations.
• Incorporating mDAs is vital for improving the accuracy of CDA prediction models.
• The study highlights the utility of heterogeneous network analysis and meta-path mining in uncovering complex biological relationships.