eTumorMetastasis

eTumorMetastasis predicts clinical outcomes from whole-exome sequencing data by converting tumor functional mutations into network-based profiles and deriving Network Operational Gene (NOG) signatures for machine-learning-based prognostic modeling.

Key Features:

• Network-based profiles: Transforms tumor functional mutations into network-based profiles representing gene interaction effects on tumor behavior.
• Network Operational Gene (NOG) signatures: Identifies NOG signatures that model the tipping point between non-recurrent and recurrent tumor states derived from genomic mutations in tumor founding clones (most recent common ancestor).
• Machine-learning predictive models: Constructs machine learning–based predictive models using network-based profiles and NOG signatures to predict clinical outcomes.
• Founder clone focus: Derives prognostic signatures specifically from mutations present in tumor founding clones (most recent common ancestor).
• Comparative performance: Demonstrates predictive accuracy that distinguishes recurred versus non-recurred breast tumors and has been reported to outperform existing genomic tests such as Oncotype DX.

Scientific Applications:

• Clinical outcome prediction: Predicts tumor recurrence and stratifies recurred versus non-recurred breast tumors from whole-exome sequencing data.
• Treatment planning guidance: Provides prognostic information derived from genomic and network features to inform patient management and therapeutic decision-making.
• Extension to other genetic diseases: The underlying network- and signature-based approach can be adapted for outcome prediction in other complex genetic diseases.

Methodology:

Transforms tumor functional mutations into network-based profiles, identifies NOG signatures from mutations in tumor founding clones (most recent common ancestor), and constructs machine learning-based predictive models using those profiles and signatures.