Detection of Copy Number Variations in Cancer Genomes from High Throughput Sequencing Data

"Copy Number estimation by a Mixture Of PoissonS" (cn.MOPS), is a well established and widely used method for detection of germline copy number variations (CNVs) in high-throughput sequencing data. cn.MOPS showed excellent performance at the detection of CNVs in HapMap samples, as well as in genomes of bacteria, fungi and plants. Since cn.MOPS constructs a model across samples for each genomic position, it is not affected by read count variations along chromosomes, and, therefore, geared to targeted sequencing. In a comparative study, cn.MOPS was the best performing method at the detection of CNVs in targeted sequencing data. However, the detection of somatic CNVs in cancer genomes is still challenging due to admixture of normal and tumor tissue, nondiploidy and very large copy number variations that affect normalization. Therefore, preprocessing, normalization, and the core algorithm of cn.MOPS have been optimized for CNV detection in cancer genomes. We demonstrate the improved performance of the enhanced cn.MOPS algorithm for cancer genomes on whole genome sequencing data from the International Cancer Genome Consortium (ICGC). cn.MOPS has been optimized for computation time and parallelized, which makes the method perfectly suited to analyze data sets of hundreds of cancer samples within a few hours.