Algorithms for detection of structural variation from NGS data rely on one or more of the following: discordant paired-end reads, split reads, or depth of coverage. Discordant paired reads are read pairs that do not map together in the ordinary way: The paired ends may map to different chromosomes or to the same chromosome either in the incorrect orientation or in the proper orientation but, for instance, too far apart in the chromosome. Split reads are single reads that map to the genome discontinuously: The first part of the read maps to one genomic region and the remainder to another. Because of the short read lengths currently available from NGS data, split reads are most useful and reliable from paired-end data, in which one end maps uniquely to the genome, serving as an “anchor,” and the other end is a split read. Finally, the depth of sequencing coverage local to a particular point in the genome provides evidence of structural variation. While changes in read depth over large regions often indicate copy number changes, more subtle variation in sequence coverage is often seen near the breakpoints of other types of structural variation.
