DWNN-RLS: regularized least squares method for predicting circRNA-disease associations

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• Circular RNAs (circRNAs) regulate gene expression and are implicated in various diseases.
• Identifying circRNA-disease associations is crucial for understanding disease mechanisms, diagnosis, and treatment.
• Wet-lab experiments for discovering these associations are costly and time-consuming.

Purpose of the Study:

• To develop an efficient computational method for predicting circRNA-disease associations.
• To overcome the limitations of experimental approaches in identifying novel circRNA-disease links.

Main Methods:

• Developed DWNN-RLS, a method utilizing Regularized Least Squares with Kronecker product kernels.
• Computed circRNA similarity using Gaussian Interaction Profile (GIP).
• Integrated disease similarity using GIP and semantic similarity (DAG).
• Employed DWNN (decreasing weight k-nearest neighbor) for initial scores and Kronecker product kernels for predictions.
• Validated performance using 5-fold, 10-fold, and leave-one-out cross-validation.

Main Results:

• DWNN-RLS achieved high AUC values: 0.8854 (5CV), 0.9205 (10CV), and 0.9701 (LOOCV).
• The method outperformed six other competing computational approaches.
• Case studies confirmed the effectiveness of DWNN-RLS in predicting novel circRNA-disease associations.

Conclusions:

• DWNN-RLS is a highly effective computational tool for predicting circRNA-disease associations.
• The method offers a significant advancement over existing computational approaches.
• This work provides a valuable resource for further research in circRNA and disease studies.