GATECDA

GATECDA predicts associations between circular RNAs (circRNAs) and drug sensitivities to identify circRNA influences on cancer progression, therapy resistance, and drug efficacy.

Key Features:

• Graph Attention Auto-Encoder (GATE): Extracts low-dimensional representations from high-dimensional, sparse features and fuses neighborhood information for circRNAs and drugs.
• Data Integration: Incorporates multiple databases including host gene sequences of circRNAs, drug structures, and known circRNA–drug sensitivity associations.
• Predictive Performance: Achieves an average Area Under the Curve (AUC) of 89.18% under 10-fold cross-validation.
• Case Studies: Demonstrates practical utility by identifying novel circRNA–drug interactions in applied analyses.

Scientific Applications:

• Oncology research: Predicts roles of circRNAs in cancer progression and drug resistance to inform experimental prioritization.
• Biomarker discovery: Identifies candidate circRNA biomarkers associated with drug response or resistance.
• Therapeutic strategy development: Supports selection of circRNA–drug interactions that may guide personalized medicine approaches.

Methodology:

Employs a Graph Attention Auto-Encoder to learn representations and fuse neighborhood information from integrated data sources (circRNA host gene sequences, drug structures, known circRNA–drug sensitivity associations) and evaluates predictive performance using 10-fold cross-validation reporting AUC.