Related Experiment Video
Updated: Jul 5, 2025
06:16
mirMachine: A One-Stop Shop for Plant miRNA Annotation
Published on: May 1, 2021
2.6K
MHCLMDA: multihypergraph contrastive learning for miRNA-disease association prediction.
Wei Peng1, Zhichen He2, Wei Dai1
1Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming, Yunnan 650500, P. R. China and Computer Technology Application Key Lab of Yunnan Province, Kunming University of Science and Technology, Kunming, Yunnan 650500, P. R. China.
Briefings in Bioinformatics
|January 20, 2024
Summary
This study introduces a new computational method, MHCLMDA, for predicting disease-associated microRNAs (miRNAs). The method improves accuracy by learning consistent feature representations across multiple views, aiding disease prevention and treatment.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Accurate prediction of microRNA (miRNA)-disease associations is crucial for disease prevention and treatment.
- Existing computational methods often overlook inter-view information interactions and feature consistency.
- This limitation hinders the effective integration of multi-view miRNA and disease data.
Purpose of the Study:
- To propose a novel computational method, Multiple Hypergraph Contrastive Learning for MiRNA-Disease Association (MHCLMDA), for predicting miRNA-disease associations.
- To address the limitations of existing methods by capturing higher-order interactions and ensuring feature consistency across multiple views.
- To enhance the accuracy and reliability of miRNA-disease association predictions.
Main Methods:
- Constructing multiple miRNA and disease hypergraphs based on similarity information.
- Applying hypergraph convolution to capture higher-order interactions within each hypergraph.
- Utilizing hypergraph contrastive learning to learn consistent miRNA and disease feature representations across different views.
- Employing a variational auto-encoder to extract features from known miRNA-disease associations.
- Fusing multi-view features for final miRNA-disease association prediction in an end-to-end manner.
Main Results:
- MHCLMDA demonstrated superior performance compared to state-of-the-art methods in predicting human miRNA-disease associations.
- The method achieved higher Area Under the Receiver Operating Characteristic Curve (AUROC) and Area Under the Precision-Recall Curve (AUPRC).
- Experimental results validate the effectiveness of the proposed hypergraph contrastive learning approach.
Conclusions:
- MHCLMDA effectively predicts miRNA-disease associations by integrating multi-view information and learning consistent feature representations.
- The proposed method offers a significant advancement over existing approaches, improving prediction accuracy.
- This work provides a valuable tool for understanding miRNA involvement in diseases and developing targeted therapies.
