CNAViz

CNAViz performs copy-number segmentation of tumor DNA sequencing data by combining local and global segmentation strategies to improve identification of copy-number aberrations (CNAs), including amplifications and deletions.

Key Features:

• Local segmentation strategy: Incorporates local genomic information to detect focal CNAs and prevent loss of small-scale events.
• Global segmentation strategy: Uses global genomic patterns to avoid overclustering and maintain consistency across larger genomic regions.
• Combined segmentation framework: Integrates local and global approaches within a user-guided framework to leverage strengths of both strategies.
• Genome partitioning: Partitions the genome into contiguous segments with uniform copy-number states for downstream analysis.
• Addresses algorithmic limitations: Targets common failures of traditional segmentation algorithms, including overclustering and missed focal events.
• Simulation benchmarking: Demonstrated improved performance over existing methods on simulated data across multiple metrics.
• Empirical validation: Validated on six bulk DNA sequencing samples from three breast cancer patients corroborated by parallel single-cell DNA sequencing data.
• Improved copy-number calling: Enhances accuracy of downstream copy-number calling from segmented tumor sequencing data.

Scientific Applications:

• CNA identification in cancer genomics: Detection and characterization of amplifications and deletions in tumor DNA sequencing data.
• Segmentation for copy-number analysis: Generation of contiguous, uniform copy-number segments for downstream analytical workflows.
• Cross-platform validation: Comparison and corroboration of bulk DNA sequencing-derived segmentation with single-cell DNA sequencing.
• Pan-cancer segmentation studies: Application to studies requiring robust segmentation and quality control of CNAs across cohorts.

Methodology:

Combines local and global segmentation strategies to partition the genome into contiguous segments with uniform copy-number states and was evaluated on simulated data as well as six bulk DNA sequencing samples from three breast cancer patients corroborated by parallel single-cell DNA sequencing.