Predicting circRNA-drug sensitivity associations by learning multimodal networks using graph auto-encoders and...

Area of Science:

• Computational biology
• Genomics
• Pharmacology

Background:

• Circular RNAs (circRNAs) expression impacts cellular drug sensitivity and overall drug efficacy.
• Experimental validation of circRNA-drug sensitivity links is resource-intensive and time-consuming.
• Developing computational tools is essential for efficient prediction of these associations.

Purpose of the Study:

• To propose MNGACDA, a novel computational method for predicting circRNA-drug sensitivity associations.
• To facilitate efficient biomedical screening by identifying potential circRNA-drug interactions.
• To address the urgent need for accurate and cost-effective prediction methods.

Main Methods:

• Constructing multimodal networks integrating circRNA and drug information.
• Utilizing node-level attention graph auto-encoders for low-dimensional embedding generation.
• Employing an inner product decoder to predict circRNA-drug sensitivity association scores.

Main Results:

• MNGACDA demonstrated superior performance compared to six state-of-the-art methods in cross-validation experiments.
• Case studies confirmed MNGACDA's effectiveness in predicting real-world circRNA-drug sensitivity associations.
• The method exhibits reliable predictive capabilities for uncovering these crucial biological relationships.

Conclusions:

• MNGACDA is a powerful and effective computational tool for predicting circRNA-drug sensitivity associations.
• The method offers a significant advancement over existing approaches, reducing experimental costs and time.
• MNGACDA's reliable predictions can guide future research in precision medicine and drug development.