CNVind

CNVind detects rare copy number variations (CNVs) from whole exome sequencing (WES) data by applying independent, region-specific depth-of-coverage normalization and segmentation to improve detection specificity.

Key Features:

• Independent Normalization Process: Performs independent depth-of-coverage normalization for each sequencing region, executing n separate normalizations when a dataset contains n sequencing regions.
• Correlation-Based Normalization: Selects the k most correlated sequencing regions using Pearson's Correlation as the distance metric and normalizes that subgroup together.
• Comprehensive Workflow: Computes depth of coverage, conducts quality control, performs independent normalization per region and combines results, then segments the normalized data and calls CNVs.
• Enhanced Specificity in Detection: Evaluation on WES data from the 1000 Genomes project reduced false positives from over 15,000 to approximately 5,000 while maintaining about 150 true positive CNVs.
• Computational Efficiency: Pipeline is optimized for cloud computing environments, computer clusters, and single-CPU servers despite the computational demands of independent normalization.

Scientific Applications:

• Complex Genetic Disorders: Detection of rare CNVs to identify candidate pathogenic variants in studies of complex genetic disorders.
• Population Genetics: Identification and characterization of rare CNVs in population-scale WES datasets.
• Personalized Medicine: Precise rare CNV detection to support genomic characterization relevant to individualized diagnostics and treatment decisions.

Methodology:

Computes depth of coverage, applies quality control, selects the k most correlated regions using Pearson's Correlation, performs independent depth-of-coverage normalization per sequencing region, combines normalized results, segments the data, and performs CNV calling.