ChimerDriver

ChimerDriver classifies gene fusions as oncogenic or non-oncogenic by integrating genetic and post-transcriptional features to assess their oncogenic potential.

Key Features:

• Neural Network-Based Classification: A neural network model integrates transcription factors, gene ontologies, microRNAs, and other functional gene information to classify fusion oncogenicity.
• Comprehensive Data Integration: Combines genetic and post-transcriptional information to distinguish oncogenic from passenger gene fusions.
• Performance Metrics: Demonstrated an F1-score of 0.83 and a recall of 96% in validation tests.
• Real-World Validation with Starfusion: Identified 21 of 24 validated gene fusion samples detected by the gene fusion detection tool Starfusion.

Scientific Applications:

• Oncogenic Fusion Assessment: Evaluates the oncogenic potential of gene fusions in cancer genomics studies.
• MicroRNA-Integrated Analysis: Incorporates microRNA-related information with genetic data to investigate post-transcriptional contributions to fusion-driven tumorigenesis.
• Diagnostic and Therapeutic Prioritization: Supports prioritization of candidate gene fusions for diagnostic evaluation and targeted therapy development.

Methodology:

A neural network integrates transcription factors, gene ontologies, microRNAs, and other functional gene features; validation was performed on samples detected with Starfusion and evaluated using F1-score and recall.