DGCLCMI: a deep graph collaboration learning method to predict circRNA-miRNA interactions.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Circular RNAs (circRNAs) function as miRNA sponges, regulating gene expression and disease.
• Wet lab methods for circRNA-miRNA association analysis are costly and time-consuming.
• Existing computational models lack deep feature extraction for circRNA-miRNA interactions, limiting prediction accuracy.

Purpose of the Study:

• To develop an efficient computational method for predicting circRNA-miRNA interactions.
• To improve the feature representation of circRNAs and miRNAs by capturing deep collaborative information.
• To overcome the limitations of current models in analyzing high-order relationships in circRNA-miRNA interactions.

Main Methods:

• Proposed a novel deep graph collaboration learning method (DGCLCMI).
• Utilized word2vec for sequence encoding into word embeddings.
• Developed a joint model combining improved neural graph collaborative filtering with a feature extraction network.

Main Results:

• DGCLCMI achieved superior performance compared to previous methods, with an average AUC of 0.960 across three datasets.
• The method effectively embeds deep interaction information into sequence representations for accurate prediction.
• A case study validated the model, with 18 out of 20 predicted unknown circRNA-miRNA interactions being accurate.

Conclusions:

• DGCLCMI enhances circRNA and miRNA feature representation through deep collaborative information.
• The model demonstrates superior predictive performance for circRNA-miRNA associations.
• DGCLCMI facilitates the discovery of novel associations and their roles in physiological processes.