A feature extraction method based on noise reduction for circRNA-miRNA interaction prediction combining...

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Circular RNAs (circRNAs) play crucial roles in biological processes by interacting with microRNAs (miRNAs).
• Understanding circRNA-miRNA interactions (CMIs) is vital for disease diagnosis and treatment.
• Existing computational methods for CMI prediction face limitations in feature extraction and efficiency, especially in sparse networks.

Purpose of the Study:

• To develop an efficient and accurate computational method for predicting circRNA-miRNA interactions (CMIs).
• To address the challenges of feature extraction in sparse networks and improve prediction performance.

Main Methods:

• Proposed JSNDCMI, a novel computational framework for CMI prediction.
• Integrated a multi-structure feature extraction framework to capture functional and topological similarities.
• Employed a Denoising Autoencoder (DAE) for robust feature representation learning.
• Utilized a Gradient Boosting Decision Tree classifier for final CMI prediction.

Main Results:

• JSNDCMI achieved superior performance in 5-fold cross-validation across all datasets.
• The method demonstrated effectiveness in predicting potential CMIs within sparse networks.
• A case study showed that 70% of the top 10 predicted CMIs were validated in PubMed.

Conclusions:

• JSNDCMI offers an effective approach for predicting circRNA-miRNA interactions.
• The developed method shows promise for advancing CMI-based diagnostics and therapeutics.
• The study provides a valuable tool for researchers in the field of RNA biology.