CNAPE

CNAPE predicts copy number alterations (CNAs) from RNA sequencing gene expression data in human cancers.

Key Features:

• Input data: Uses gene expression data derived from RNA sequencing (RNA-seq) as the input signal for CNA inference.
• Prediction scope: Infers DNA copy number variations across entire chromosomes, chromosomal arms, and specific cancer-associated genes.
• Model architecture: Employs a machine learning model informed by prior knowledge.
• Training and validation: Model was trained and validated on 9,740 cancer samples from The Cancer Genome Atlas (TCGA).
• Performance: Achieves over 80% accuracy for broad genomic regions and demonstrates comparable precision in certain focal areas.
• Demonstrated use case: Application has been demonstrated in gliomas.

Scientific Applications:

• Inference when DNA is scarce: Infers CNAs in contexts where high-quality DNA is unavailable, including biopsies and single-cell genomes that lack microgram-level DNA.
• Cancer genomics mapping: Maps CNAs across chromosomes, chromosomal arms, and focal genes in human cancer studies.
• Glioma research: Supports investigation of copy number alterations in gliomas.

Methodology:

Predicts CNAs by applying a prior-knowledge-informed machine learning model to RNA-seq gene expression data, with training and validation performed on 9,740 TCGA cancer samples.