Predicting circRNA-Disease Associations Based on Deep Matrix Factorization with Multi-source Fusion.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Circular RNAs (circRNAs) are increasingly recognized for their roles in disease pathogenesis.
• Experimental methods for identifying circRNA-disease associations are costly and time-consuming.
• There is a critical need for efficient computational tools to predict these associations.

Purpose of the Study:

• To develop and validate a novel computational approach for predicting circRNA-disease associations.
• To leverage multi-source data and advanced matrix factorization techniques for improved prediction accuracy.
• To provide an efficient alternative to experimental methods in circRNA research.

Main Methods:

• Developed a Deep Matrix Factorization with Multi-Source Fusion (DMFMSF) model.
• Integrated diverse circRNA and disease similarity measures using similarity kernel fusion.
• Employed Singular Value Decomposition (SVD) and deep matrix factorization to capture linear and non-linear features.
• Validated performance using Leave-One-Out Cross-Validation (LOOCV) and 5-fold Cross-Validation (5-fold CV) on benchmark datasets.

Main Results:

• DMFMSF demonstrated superior performance compared to existing computational methods.
• The approach effectively predicted potential circRNA-disease associations.
• Case studies on five major diseases (hepatocellular carcinoma, breast cancer, acute myeloid leukemia, colorectal cancer, coronary artery disease) showed promising results.

Conclusions:

• DMFMSF is an accurate and efficient computational tool for predicting circRNA-disease associations.
• The method offers a valuable resource for researchers investigating the roles of circRNAs in disease.
• This approach can accelerate the discovery of novel circRNA-disease links, aiding in disease diagnosis and treatment strategies.