CONDEX: COpy Number Detection in EXome Sequences

Exome sequencing by hybrid capture facilitates obtaining cost-effective, comprehensive data on coding sequence variation from short reads. Standard analysis tools focus on detecting and characterizing variants of single or a few nucleotides while copy number variants (CNVs) that span multiple regions of exon baits have not yet been considered. Here, we present a Hidden Markov Model based method to identify CNVs from exome sequencing data. Using depth coverage and the heterozygosity of SNPs, we call CNVs with per-exon training data from other samples. The method has >90% accuracy in identification of deletions and insertions in simulations.