A novel circRNA-miRNA association prediction model based on structural deep neural network embedding

Area of Science:

• Bioinformatics
• Molecular Biology
• Genomics

Background:

• Circular RNAs (circRNAs) play significant roles in complex diseases by regulating microRNA (miRNA) target genes.
• Traditional experimental methods for identifying circRNA-miRNA associations are labor-intensive, costly, time-consuming, and inefficient.

Purpose of the Study:

• To develop a novel computational model for predicting potential circRNA-miRNA associations.
• To offer a more efficient and accurate alternative to experimental approaches in identifying circRNA-miRNA interactions.

Main Methods:

• Proposed a computational model named WSCD (Word2vec, SDNE, Convolutional Neural Network, and Deep Neural Network).
• Utilized word embedding (Word2vec) and graph embedding (SDNE) to extract attribute and behavior features.
• Integrated these features into a fusion model combining Convolutional Neural Network and Deep Neural Network to predict circRNA-miRNA interactions.

Main Results:

• Achieved a prediction accuracy of 81.61% and an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.8898.
• Demonstrated significantly higher accuracy compared to state-of-the-art and traditional classifier models.
• Validated 23 out of the top 30 predicted miRNA-related circRNAs in experimental settings.

Conclusions:

• The WSCD model provides a reliable and supplementary computational method for predicting potential miRNA-circRNA associations.
• WSCD offers a more efficient and cost-effective approach compared to traditional wet laboratory experiments.
• This advancement aids in understanding the role of circRNA-miRNA interactions in complex diseases.