CNVnator

CNVnator detects and genotypes copy number variations (CNVs) from whole-genome sequencing data using read-depth analysis.

Key Features:

• Read-depth analysis: Uses read-depth from personal/whole-genome sequencing to discover and genotype CNVs.
• Mean-shift segmentation: Employs a refined mean-shift approach with multiple-bandwidth partitioning to identify CNV regions.
• GC-content correction: Applies GC content correction to adjust read-depth signals for sequence composition bias.
• Calibration and validation: Calibrated and validated using validation data from the 1000 Genomes Project.
• Performance metrics: Reports sensitivity of 86%–96%, false-discovery rate of 3%–20%, genotyping accuracy of 93%–95%, and breakpoint resolution typically <200 base pairs in 90% of cases with sufficient sequencing coverage.
• Complementary detection: Identifies over half of validated CNVs missed by split-read and read-pair methods while not detecting CNVs generated by retrotransposable elements.
• Population genotyping: Genotypes CNVs across CEPH, Yoruba, and Chinese-Japanese cohorts and identifies multi-allelic loci.
• De novo and multi-allelic CNV detection: Detects de novo and multi-allelic CNVs and contributed to the identification of six potential de novo CNVs in two family trios.

Scientific Applications:

• Population studies: Characterizes CNV distributions and genetic diversity within and between populations.
• Characterization of atypical CNVs: Detects de novo and multi-allelic CNVs and complex loci for downstream analysis.
• Genetic research: Enables genotyping across CEPH, Yoruba, and Chinese-Japanese groups and supports estimates that at least 11% of CNV loci are multi-allelic.
• Evolutionary analysis: Detects deviations from Hardy-Weinberg equilibrium at complex loci, indicating potential selection pressures.
• Familial studies: Supports discovery of potential de novo CNVs within family trios for hereditary investigations.

Methodology:

Performs read-depth analysis using a refined mean-shift algorithm with multiple-bandwidth partitioning and GC-content correction, calibrated and validated against 1000 Genomes Project data, and reports breakpoint localization typically <200 base pairs with sufficient sequencing coverage.