CNAtra

CNAtra detects and classifies copy number alterations (CNAs) in cancer genomes from low-coverage whole-genome sequencing (WGS) data to enable identification of both large-scale and focal structural copy-number changes.

Key Features:

• Hierarchical detection framework: Distinguishes large-scale and focal CNAs using a hierarchical approach tailored to multi-scale aberrations in cancer genomes.
• Single-sample CNV profiling: Performs copy-number profiling from single WGS samples using read depth (RD) signals.
• Multimodal reference estimation: Estimates the reference copy number from complex RD profiles using a multimodal distribution.
• Signal processing and segmentation: Applies Savitzky-Golay filtering and Modified Varri segmentation to detect change points in RD signals.
• CN state-driven merging algorithm: Merges segments based on inferred copy-number states to identify large segments with distinct copy numbers.
• Coverage-based thresholding for focal alterations: Uses coverage-based thresholding within each identified large segment to pinpoint focal alterations and reduce false positives.
• Optimization for low-coverage data: Designed to maintain sensitivity for both focal and large-scale alterations in low-coverage WGS datasets.

Scientific Applications:

• Cancer genomics research: Supports analysis of genome evolution and pathological mechanisms by accurately detecting both large-scale and focal CNAs in cancer genomes.
• Benchmarking and validation: Validated with experimentally verified segmental aneuploidies and focal alterations and benchmarked on simulated data, reporting sensitivities of 93% for focal and 97% for large-scale alterations and superior precision, recall, and F-measure relative to other tools.

Methodology:

Models read-depth (RD) signals with a multimodal distribution, applies Savitzky-Golay filtering and Modified Varri segmentation to find change points, employs a CN state-driven merging algorithm to define large segments, and uses coverage-based thresholding within segments to detect focal CNAs.