Fig. 2: Landscape of genomic structural variation.

a, Number of non-redundant variants identified across the cohort (n = 141). Large indels (left; 20–49 bp) and SVs (right; ≥50 bp) are shown separately and parsed by type: non-repetitive, tandem repeats (homopolymer (HOMO), STR and TR) and mobile elements (long terminal repeat (LTR), long interspersed nuclear elements (LINE), short interspersed nuclear element (SINE), retroposon and DNA/DNA transposon (DNA)). Light shades represent deletions and dark shades represent insertions. b, Frequency of non-redundant SVs relative to distance from the nearest telomere. c, Size distribution of insertions (positive values) and deletions (negative values), parsed by type (colour scheme as in a). Characteristic peaks for Alu elements (280 bp) and L1 elements (6 kb) are marked. d, Size distributions for each variant type. Mobile element SVs are classified as ‘complete’ if they encompass one or more complete annotated element or ‘fragment’ if they encompass only part of an annotated element. e, Number of non-redundant SVs found in a search performed against a combination of the gnomAD, deCODE and HGSVC (freeze 4) SV annotations. SVs were first lifted from T2T-chm13 to hg38. Some could not be lifted because they were deleted or partially deleted in hg38. SVs that could be lifted were categorized as ‘annotated’ or ‘unannotated’ on the basis of reciprocal overlap to any single annotated SV. High (>80%) and moderate (50–80%) overlaps were considered as annotated, whereas low (<50%; beige) and no overlap were considered as unannotated.