The performance of any method for detection of structural variation depends critically on the type of sequencing data available. For instance, split-read methods to detect trans-locations generally require adequate coverage so that the translocation breakpoints are spanned by several split reads, and they will not perform well using low coverage whole genome sequencing data. Similarly, indels can be detected from exome (or targeted-capture) data using paired- or split-read methods only if at least one of the breakpoints falls within or near the captured regions. Finally, any method that relies on read depth will perform differently for whole genome as compared with exome or targeted-capture data, as the depth of coverage in targeted-capture data is particularly susceptible to GC bias, uneven coverage near the boundaries of the capture baits, and other systematic biases. A summary of tools used for structural variation detection is presented in Table 1.
