Related Experiment Video
Updated: May 31, 2025
Characterization of Functionally Associated miRNAs in Glioblastoma and their Engineering into Artificial Clusters for Gene Therapy
Published on: October 4, 2019
Graph Convolutional Network with Neural Collaborative Filtering for Predicting miRNA-Disease Association
1Major of Big Data Convergence, Division of Data Information Science, Pukyong National University, Busan 48513, Republic of Korea.
Abstract:
Background: Over the past few decades, micro ribonucleic acids (miRNAs) have been shown to play significant roles in various biological processes, including disease incidence. Therefore, much effort has been devoted to discovering the pivotal roles of miRNAs in disease incidence to understand the underlying pathogenesis of human diseases. However, identifying miRNA-disease associations using biological experiments is inefficient in terms of cost and time. Methods: Here, we discuss a novel machine-learning model that effectively predicts disease-related miRNAs using a graph convolutional neural network with neural collaborative filtering (GCNCF). By applying the graph convolutional neural network, we could effectively capture important miRNAs and disease feature vectors present in the network while preserving the network structure. By exploiting neural collaborative filtering, miRNAs and disease feature vectors were effectively learned through matrix factorization and deep learning, and disease-related miRNAs were identified. Results: Extensive experimental results based on area under the curve (AUC) scores (0.9216 and 0.9018) demonstrated the superiority of our model over previous models. Conclusions: We anticipate that our model could not only serve as an effective tool for predicting disease-related miRNAs but could be employed as a universal computational framework for inferring relationships across biological entities.
Insights
This study introduces a new machine learning model, GCNCF, to efficiently predict micro ribonucleic acid (miRNA) and disease associations. The model significantly outperforms previous methods, offering a faster and more cost-effective approach to identifying disease-related miRNAs.
Area of Science:
- Genomics
- Computational Biology
- Bioinformatics
Background:
- Micro ribonucleic acids (miRNAs) are crucial regulators in biological processes and disease development.
- Identifying miRNA-disease associations is vital for understanding human disease pathogenesis.
- Experimental methods for miRNA-disease association discovery are time-consuming and costly.
Purpose of the Study:
- To develop an efficient computational model for predicting miRNA-disease associations.
- To overcome the limitations of experimental approaches in identifying these relationships.
Main Methods:
- A novel machine learning model, Graph Convolutional Neural Network with Neural Collaborative Filtering (GCNCF), was developed.
- GCNCF utilizes graph convolutional networks to capture miRNA and disease feature vectors.
- Neural collaborative filtering is employed for effective feature learning through matrix factorization and deep learning.
Main Results:
- The GCNCF model demonstrated superior performance in predicting miRNA-disease associations.
- Area under the curve (AUC) scores of 0.9216 and 0.9018 validated the model's effectiveness.
- The model significantly outperformed existing methods in experimental evaluations.
Conclusions:
- The GCNCF model provides an effective computational tool for predicting disease-related miRNAs.
- This framework can be broadly applied to infer relationships between various biological entities.
- The study highlights the potential of machine learning in accelerating biological discovery.
