KS-CMI: A circRNA-miRNA interaction prediction method based on the signed graph neural network and denoising...

Area of Science:

• Biochemistry
• Genomics
• Computational Biology

Background:

• Circular RNAs (circRNAs) are vital biomarkers for human disease diagnosis, treatment, and prognosis.
• Identifying circRNA-microRNA interactions (CMI) guides crucial biological experiments.
• Current CMI prediction models face limitations due to scarce experimental data and high randomness.

Purpose of the Study:

• To develop a novel and effective method, KS-CMI, for predicting circRNA-miRNA interactions (CMI) in real-world cases.
• To enhance the accuracy and reliability of CMI prediction beyond existing computational approaches.

Main Methods:

• Constructed circRNA-miRNA-cancer (CMCI) networks to enrich molecular 'behavior relationships'.
• Extracted molecular behavior attributes using balance theory.
• Employed a denoising autoencoder (DAE) for enhanced molecular feature representation.
• Utilized the CatBoost classifier for CMI prediction.

Main Results:

• KS-CMI demonstrated highly reliable prediction results in real-world applications.
• The method achieved competitive performance across all tested CMI prediction datasets.
• The approach effectively addresses the limitations of existing CMI prediction models.

Conclusions:

• KS-CMI offers a robust and effective solution for predicting circRNA-miRNA interactions.
• The method's performance in real cases signifies its practical utility in biomedical research.
• This advancement facilitates more accurate CMI prediction, supporting future biological investigations.