Related Experiment Video
Updated: Jul 26, 2025
07:35
A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports
Published on: October 13, 2023
1.7K
MPCLCDA: predicting circRNA-disease associations by using automatically selected meta-path and contrastive learning
Wei Liu1, Ting Tang1, Xu Lu2,3
1School of Computer Science, Xiangtan University, Xiangtan, 411105, China.
Briefings in Bioinformatics
|June 16, 2023
Summary
This study introduces MPCLCDA, a novel computational model for predicting circular RNA-disease associations (CDAs). The model effectively identifies potential CDAs, aiding disease understanding and treatment by overcoming data limitations.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Circular RNAs (circRNAs) are implicated in human diseases, making their association identification crucial for medical advancements.
- Traditional methods for identifying circRNA-disease associations (CDAs) are inefficient and labor-intensive.
- Existing computational models face challenges due to limited data, high dimensionality, and data imbalance.
Purpose of the Study:
- To develop an advanced computational model for accurately predicting circRNA-disease associations (CDAs).
- To address data limitations and improve the efficiency of CDA prediction.
- To enhance disease prevention, diagnosis, and treatment strategies through reliable circRNA-disease linkage.
Main Methods:
- Constructed a heterogeneous network integrating circRNA similarity, disease similarity, and known associations.
- Employed automatically selected meta-paths and graph convolutional networks to derive low-dimensional node features.
- Utilized contrastive learning to optimize node features for clearer distinction between positive and negative samples.
- Predicted circRNA-disease scores using a multilayer perceptron.
Main Results:
- The MPCLCDA model achieved high predictive performance across four datasets.
- Average performance metrics included Area Under the Receiver Operating Characteristic Curve (AUC) of 0.9752, Area Under the Precision-Recall Curve (AUPRC) of 0.9831, and F1 score of 0.9745.
- Case studies validated the model's predictive capability and practical application in understanding human diseases.
Conclusions:
- The proposed MPCLCDA model offers a powerful and efficient approach for predicting circRNA-disease associations.
- This method effectively overcomes data limitations inherent in previous computational models.
- The findings highlight the potential of MPCLCDA in advancing circRNA-related disease research and clinical applications.
