KGDCMI: A New Approach for Predicting circRNA-miRNA Interactions From Multi-Source Information Extraction and Deep...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Circular RNAs (circRNAs) act as microRNA (miRNA) sponges, regulating gene expression.
• Understanding circRNA-miRNA interactions is crucial for diagnosing and treating complex diseases.
• Traditional experimental methods for predicting these interactions are costly and time-consuming.

Purpose of the Study:

• To develop an effective computational method for predicting circRNA-miRNA interactions.
• To address the limitations of existing computational models in this field.

Main Methods:

• Proposed KGDCMI, a novel computing method integrating multi-source information.
• Extracted RNA attribute information from sequence and similarity data.
• Utilized graph-embedding algorithms, principal component analysis, and deep neural networks for feature extraction, fusion, and prediction.

Main Results:

• KGDCMI achieved high prediction accuracy with AUC = 89.30% and AUPR = 87.67%.
• Outperformed the existing model by 2.37% (AUC) and 3.08% (AUPR).
• Case study validated 70% of the top 10 predicted interactions in PubMed.

Conclusions:

• KGDCMI is a feasible and effective computational tool for predicting circRNA-miRNA interactions.
• The method offers a reliable and cost-efficient approach for biological experiments.
• KGDCMI advances the understanding of RNA interactions in disease pathogenesis.