Related Experiment Video
Updated: Sep 12, 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
GPLMD: A Multi-Task Graph Learning Framework for inferring the relationships among lncRNAs, miRNAs and diseases
Rong Sun1, Xun Chen2, Dan Zhao3
1School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, 611731, Sichuan, China.
Computers in Biology and Medicine
|August 10, 2025
Summary
This study introduces GPLMD, a novel computational model for predicting relationships between long non-coding RNAs (lncRNAs), microRNAs (miRNAs), and diseases. GPLMD effectively integrates multiple graph learning techniques to enhance disease prediction and identify novel therapeutic targets.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) are critical in disease pathogenesis, influencing prevention, diagnosis, and treatment.
- Experimental validation of lncRNA-miRNA-disease associations is costly and time-consuming, necessitating efficient computational approaches.
- Existing computational methods often fail to integrate diverse biological data or leverage shared knowledge across related prediction tasks.
Purpose of the Study:
- To develop a novel computational model, GPLMD, for inferring complex relationships among lncRNAs, miRNAs, and diseases.
- To address limitations of single-task prediction and enhance the modeling of intricate biological interactions from diverse graph structures.
- To improve the accuracy and efficiency of predicting lncRNA-disease associations (LDA), miRNA-disease associations (MDA), and lncRNA-miRNA interactions (LMI).
Main Methods:
- Constructed multiple graph types: bipartite, feature structural, and meta-path graphs, capturing biomolecular connections.
- Employed a graph convolutional network (GCN)-based decoder for learning node topology and an attention-based approach for integrating diverse graph views.
- Integrated convolutional neural networks (CNNs) with graph learning embeddings for joint optimization in classification tasks.
Main Results:
- GPLMD significantly outperformed baseline methods in predicting lncRNA-disease associations (LDA), miRNA-disease associations (MDA), and lncRNA-miRNA interactions (LMI) on two benchmark datasets.
- Case studies validated GPLMD's capability in identifying novel disease-related lncRNAs and miRNAs, demonstrating its practical utility.
- The model successfully integrated multi-type neighbor topology and rich global representations for accurate relationship inference.
Conclusions:
- GPLMD offers a powerful and integrated approach for predicting lncRNA-miRNA-disease associations, overcoming limitations of existing methods.
- The model's ability to leverage diverse biological data and complex graph structures enhances prediction accuracy.
- GPLMD holds promise for accelerating the discovery of novel biomarkers and therapeutic targets in disease research.
