GMNN2CD: identification of circRNA-disease associations based on variational inference and graph Markov neural...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Circular RNAs (circRNAs) are increasingly recognized for their crucial roles in various diseases.
• Understanding circRNA-disease relationships is vital for disease pathogenesis and therapeutic strategies.
• Traditional biotechnological methods for identifying these associations are often inefficient.

Purpose of the Study:

• To develop an efficient computational method for predicting novel circRNA-disease associations.
• To leverage graph Markov neural networks (GMNN) for enhanced association prediction.

Main Methods:

• GMNN2CD utilizes verified circRNA-disease associations to compute semantic and Gaussian interactive profile kernel similarities.
• A fusion feature variational map autoencoder learns deep features, while a label propagation map autoencoder propagates known associations.
• Variational inference and GMNN alternate training optimize high-dimensional feature extraction from low-dimensional representations.

Main Results:

• GMNN2CD demonstrated superior performance compared to state-of-the-art methods across five benchmark datasets via 5-fold cross-validation.
• Case studies confirmed GMNN2CD's capability in detecting potential circRNA-disease associations.
• The method effectively predicts unknown circRNA-disease associations.

Conclusions:

• GMNN2CD offers an efficient and accurate computational approach for circRNA-disease association prediction.
• This method can significantly contribute to understanding disease mechanisms and identifying therapeutic targets.
• The developed tool and data are publicly available for further research.