Prediction of the Association between Transfer RNA and Diseases: A Deep Learning Approach Combining Multi-View Graph...

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Identifying transfer RNA (tRNA) and disease associations is crucial for medical advancements.
• Computational approaches provide an efficient pathway for discovering these complex relationships.

Purpose of the Study:

• To develop and validate an advanced computational model for predicting tRNA-disease associations.
• To leverage network-based methods and deep learning for enhanced biological insights.

Main Methods:

• Proposed the MGC2ATDA model, integrating a multiview graph convolutional network, scaled attention fusion, and cross-attention.
• Constructed a multiview network using tRNA sequence data, disease similarity, and known associations.
• Employed graph convolutional networks and attention mechanisms for feature extraction and prediction.

Main Results:

• MGC2ATDA achieved high performance (AUC 0.8786, AUPR 0.3657) on tRNA-disease data, surpassing existing methods.
• Demonstrated strong generalization on piRNA-disease data (AUC 0.9353, AUPR 0.6105).
• Ablation studies confirmed the effectiveness of the scaled attention fusion module.

Conclusions:

• MGC2ATDA is an efficient and accurate computational tool for identifying potential tRNA-disease associations.
• The model offers significant potential for advancing biomedical research and therapeutic development.