HECLCDA:CircRNA-Drug Sensitivity Prediction via Heterogeneous Cross-Scale Contrastive Learning.

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Circular RNAs (circRNAs) are crucial non-coding RNAs involved in cancer development and drug resistance.
• Traditional experimental methods for identifying circRNA-drug sensitivity links are inefficient and costly.
• Accurate computational tools are needed to predict circRNA-drug sensitivity associations.

Purpose of the Study:

• To develop an efficient and accurate computational method for predicting circRNA-drug sensitivity associations.
• To address the limitations of traditional experimental verification methods.
• To facilitate the discovery of novel therapeutic strategies.

Main Methods:

• Proposed HECLCDA, a novel method based on heterogeneous cross-scale contrastive learning.
• Integrated circRNA sequence similarity, drug structural similarity (SIS), and Gaussian kernel similarity.
• Constructed a heterogeneous graph and employed a Heterogeneous Graph Transformer for topological information extraction.
• Introduced a cross-scale contrastive learning mechanism for enhanced node embedding discrimination.

Main Results:

• HECLCDA demonstrated excellent performance on real data.
• The method efficiently predicts drug sensitivity.
• Cross-validation and case studies validated the model's effectiveness in identifying potential circRNA-drug sensitivity associations.

Conclusions:

• HECLCDA offers an efficient and accurate computational approach for predicting circRNA-drug sensitivity.
• The developed method can accelerate the discovery of novel circRNA-drug interactions.
• This tool holds promise for advancing cancer therapy and personalized medicine.