Router-guided dual attention with graph masked autoencoder for microRNA-drug association prediction

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• MicroRNAs (miRNAs) are crucial regulators of gene expression and key players in drug response.
• Accurate identification of miRNA-drug associations is vital for understanding drug mechanisms and advancing therapeutics.
• Current computational methods face challenges in capturing both global and local patterns in biological data, especially with sparse or noisy information.

Purpose of the Study:

• To develop an advanced computational framework for predicting miRNA-drug associations.
• To overcome limitations of existing methods in modeling complex dependencies and handling noisy biological data.
• To enhance the discovery of potential therapeutic targets and drug-response biomarkers.

Main Methods:

• Introduced router-guided dual attention with graph masked autoencoder (RGDAGMAE).
• Employed a router-guided dual attention module for efficient global dependency modeling.
• Utilized a gated graph masked autoencoder for self-supervised local structure reconstruction and noise suppression.

Main Results:

• RGDAGMAE demonstrated superior performance over existing methods on three benchmark datasets.
• Ablation studies validated the effectiveness of adaptive attention routing and masked graph reconstruction.
• Case studies highlighted the biological relevance of predicted miRNA-drug associations, such as with vorinostat and miR-509-3p.

Conclusions:

• RGDAGMAE offers a robust and effective approach for predicting miRNA-drug associations.
• The framework successfully integrates global and local pattern recognition for improved accuracy.
• Findings support RGDAGMAE's utility in prioritizing candidate miRNA-drug associations for further investigation in drug discovery.