Generative Adversarial Matrix Completion Network based on Multi-Source Data Fusion for miRNA-Disease Associations...

Area of Science:

• Biomedical Informatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are crucial in complex diseases, serving as potential biomarkers and therapeutic targets.
• Computational methods are increasingly used to identify disease-associated miRNAs.
• Existing models face limitations due to insufficient data fusion and incomplete association knowledge.

Purpose of the Study:

• To develop an advanced computational model for predicting miRNA-disease associations.
• To overcome limitations of previous models in data fusion and handling incomplete information.
• To enhance the accuracy and reliability of identifying disease-related miRNAs.

Main Methods:

• Proposing Generative Adversarial Matrix Completion Network based on Multi-source Data Fusion (GAMCNMDF).
• Integrating diverse data sources with a nonlinear fusion approach to update miRNA and disease similarity networks.
• Utilizing a 'hint' mechanism to enable successful predictions with incomplete data.

Main Results:

• GAMCNMDF demonstrated superior performance in 10-fold cross-validation on two databases.
• The model achieved outstanding results in identifying small molecule-related miRNAs.
• Case studies on neoplasms confirmed GAMCNMDF as a promising prediction tool.

Conclusions:

• GAMCNMDF offers a robust and effective approach for miRNA-disease association prediction.
• The model's ability to integrate diverse data and handle incomplete information enhances its applicability.
• GAMCNMDF shows significant potential for advancing disease diagnosis and therapeutic target identification.