cnvOffSeq

cnvOffSeq detects intergenic copy number variation (CNV) by exploiting off-target read depth from whole-exome sequencing data to enable CNV detection in non-coding regions.

Key Features:

• Local adaptive SVD normalization: Uses a novel normalization framework based on local adaptive singular value decomposition (SVD) to model and correct heterogeneous off-target coverage.
• Benchmark performance: Validated on whole-exome sequencing samples from the 1000 Genomes Project with sensitivity 57.5%, specificity 99.2% and false discovery rate (FDR) 5% for 104 gold-standard intergenic deletions.
• High sensitivity for larger events: Achieves 90.4% sensitivity for deletions larger than 5 kilobases while maintaining low FDR.
• Improved accuracy over alternatives: Outperforms existing whole-genome and whole-exome CNV detection methods and improves on naïve local SVD approaches for intergenic CNV detection.

Scientific Applications:

• Intergenic CNV discovery: Enables detection and genotyping of copy number variations in non-coding, intergenic regions using off-target exome reads.
• Genetics research beyond exons: Extends whole-exome sequencing studies to survey genomic variation outside protein-coding regions for studies of complex traits and disease-associated non-coding variation.

Methodology:

Normalization of off-target read depth via local adaptive SVD followed by CNV detection and genotyping using the normalized depth signal.