exomeCopy

exomeCopy detects copy number variants (CNVs) from exome sequencing samples, including unpaired samples, by modeling and segmenting read depth while correcting for positional covariates such as GC-content and background read depth.

Key Features:

• Hidden Markov Model: exomeCopy applies a hidden Markov model to raw read count data to infer CNV states.
• Positional covariates: The method incorporates GC-content and other positional covariates to normalize variation in read depth.
• Background read depth from control sets: It leverages background read depth derived from control exome samples to improve CNV detection relative to a reference.
• Simultaneous normalization and segmentation: exomeCopy normalizes and segments samples concurrently to identify regions with constant copy counts.
• Support for unpaired samples: The approach detects CNVs in unpaired exome sequencing samples without requiring matched controls.

Scientific Applications:

• Chromosome X exome projects: Detection of numerous unique CNVs in large-scale chromosome X exome sequencing studies.
• Cross-platform validation: Validation of predictions using cross-platform control sets from publicly available exome sequencing data.

Methodology:

exomeCopy applies a hidden Markov model to raw read counts and simultaneously normalizes and segments samples by integrating background read depth and positional covariates (e.g., GC-content); simulation studies were used to evaluate sensitivity for heterozygous and homozygous CNVs.