NCMD: Node2vec-Based Neural Collaborative Filtering for Predicting MiRNA-Disease Association

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are crucial in disease pathogenesis due to their role in regulating gene expression.
• Identifying disease-associated miRNAs is vital for developing prognostic markers and therapeutic strategies.
• Computational methods are increasingly important for miRNA-disease association prediction due to the limitations of experimental approaches.

Purpose of the Study:

• To propose a novel computational framework, NCMD, for predicting miRNA-disease associations.
• To leverage deep learning techniques, specifically Node2vec and neural collaborative filtering, for enhanced prediction accuracy.
• To validate the effectiveness of NCMD in identifying disease-related miRNAs.

Main Methods:

• Node2vec was employed to learn low-dimensional vector representations for miRNAs and diseases.
• A deep learning model combining generalized matrix factorization and a multilayer perceptron was utilized.
• The framework, NCMD, was developed for predicting miRNA-disease associations.

Main Results:

• NCMD demonstrated comparable performance to state-of-the-art methods based on statistical evaluations.
• Case studies involving breast, lung, and pancreatic cancers validated the predictive capabilities of NCMD.
• The study highlighted the benefits of using a neural collaborative filtering approach for discovering novel miRNA-disease links.

Conclusions:

• NCMD offers a powerful and effective computational approach for predicting miRNA-disease associations.
• The framework has the potential to accelerate the discovery of novel biomarkers and therapeutic targets.
• Deep learning-based neural collaborative filtering is a promising strategy for advancing miRNA research.