Copy number variant detection with low-coverage whole-genome sequencing is a viable alternative to the traditional array-CGH

Copy number variations (CNVs) are a type of structural variants involving alterations in the number of copies of specific regions of DNA, which can either be deleted or duplicated. CNVs contribute substantially to normal population variability; however, abnormal CNVs cause numerous genetic disorders. Nowadays, several methods for CNV detection are used, from the conventional cytogenetic analysis through microarray-based methods (aCGH) to next-generation sequencing (NGS). We present GenomeScreen – NGS-based CNV detection method for lowcoverage whole-genome sequencing. We determined the theoretical limits of its accuracy and confirmed it with extensive in-silico study and real patient samples with known genotypes. Theoretically, at least 6M uniquely mapped reads are required to detect CNV with a length of 100 kilobases (kb) or more with high confidence (Z-score > 7). In practice, the in-silico analysis showed the requirement of at least 8M to obtain >99% accuracy (for 100 kb deviations). We compared GenomeScreen with one of the currently used aCGH methods in diagnostic laboratories, which has a 200 kb mean resolution. GenomeScreen and aCGH both detected 59 deviations, GenomeScreen furthermore detected 134 other (usually) smaller variations. The performance of the proposed GenemoScreen tool is comparable or superior to the aCGH regarding accuracy, turnaround time, and cost-effectiveness, presenting a reasonable benefit particularly in a prenatal diagnosis setting.