Computational Prediction of Human Disease- Associated circRNAs Based on Manifold Regularization Learning Framework

Area of Science:

• Genomics and Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• Circular RNAs (circRNAs) play vital roles in biological processes, including tumorigenesis.
• The functions of most circRNAs are unknown, and large-scale discovery of disease-associated circRNAs is limited.
• Omics data and computational models offer opportunities to identify disease-associated circRNAs.

Purpose of the Study:

• To develop a computational framework (MRLDC) for large-scale identification of disease-associated circRNAs.
• To explore the relationship between circRNA function and disease association.
• To aid in understanding complex disease pathogenesis at the circRNA level.

Main Methods:

• Computed Gaussian interaction profile kernel similarity for circRNAs and diseases.
• Constructed a heterogeneous circRNA-disease bilayer network.
• Developed a weighted low-rank approximation optimization algorithm with dual-manifold regularizations.

Main Results:

• The MRLDC framework effectively identifies candidate disease-associated circRNAs with high accuracy.
• Experimental results validate the method's predictive capability.
• Case studies demonstrate MRLDC's potential in discovering novel circRNA-disease associations.

Conclusions:

• MRLDC provides an effective computational approach for identifying disease-associated circRNAs.
• This method can advance the exploration of circRNA functions in disease.
• Findings support potential applications in disease diagnosis and therapeutic strategies.