Discovering Patterns of Structural Variation by Mining Molecular Fossils

mation thus obtained allows inferring the possible mechanism(s) of origin of the SVs [Kloosterman et al., 2012; Poot and Haaf, 2015]. Since these analyses are laborious, they have been performed only on a case-by-case basis for selected individuals, which limits the number of cases analyzed and may have introduced ascertainment biases. Recently, a converse approach has been taken. SVs were identified by comparing sequence data from whole genome sequencing projects of healthy individuals derived from the general population with the human reference genome sequence. All deviations from the human reference genome sequence can be considered as ‘molecular fossils’ representing SVs. These studies not only provided unprecedented insights into the mechanism(s) of origin of CNVs, inversions, and indels, but also an estimate of their relative frequency in the general population. To distinguish nonallelic homologous recombination (NAHR), with long stretches of homology surrounding the breakpoints, from transposable element insertions (TEI), with short homologies containing mobile elements close to the SV, from nonhomologous end-joining (NHEJ), with little or no homology at breakpoints, and from template-switching mechanisms during replication (FoSTeS), Abyzov et al. [2015] analyzed 8,943 deletion breakpoints in 1,092 whole genome sequences. All breakpoints were associated with evolutionarily less-conserved regions, spanning up to hundreds of kilobases downstream and upstream from the breakpoints of the SVs. TEI breakpoints were 5 times more likely to reside in hyStructural genome variations (SVs) are defined as changes in the organization of DNA sequence elements involving at least 50 bp, while smaller structural genome changes are known as indels [Alkan et al., 2011]. Conventionally, SVs are classified according to the number of chromosome breaks by which they originated as either copy number variations (CNVs) consisting of terminal deletions and duplications (1-break events), interstitial deletions and duplications (2-break events), or as complex chromosome rearrangements involving more than 2 breaks [Poot and Haaf, 2015]. CNVs larger than 100 kb arise de novo at an estimated rate of ∼ 1.2 × 10 –2 CNVs per meiosis [Itsara et al., 2010]. All types of germline SVs together occur more frequently than germline single nucleotide variations (SNVs), affect more nucleotides, and may have a greater phenotypic impact than SNVs [Stankiewicz and Lupksi, 2010; Campbell and Eichler, 2013]. In up to 20% of children with developmental and/ or intellectual delay, phenotypically significant SVs can be detected upon investigation by both classical karyotyping and array CGH [Hochstenbach et al., 2009; Cooper et al., 2011]. This underscores the need for comprehensive analyses of SVs in the human genome. Thousands of human genomes have been analyzed by classical karyotyping, FISH, and genome-wide array CGH. While this provided information regarding the structure of the SVs, their breakpoints could not be resolved at nucleotide resolution. For the latter, paired-end and mate-pair sequencing has been used [Kloosterman et al., 2012]. The inforAccepted: September 12, 2016 by M. Schmid Published online: October 21, 2016