Structural mosaic abnormalities are large post-zygotic mutations present in a subset of cells and have been implicated in developmental disorders and cancer. Such mutations have been conventionally assessed in clinical diagnostics using cytogenetic or microarray testing. Modern disease studies rely heavily on exome sequencing, yet an adequate method for the detection of structural mosaicism using targeted sequencing data is lacking. Here, we present a method, called MrMosaic, to detect structural mosaic abnormalities using deviations in allele fraction and read coverage from next-generation sequencing data. Whole-exome sequencing (WES) and whole-genome sequencing (WGS) simulations were used to calculate detection performance across a range of mosaic event sizes, types, clonalities, and sequencing depths. The tool was applied to 4911 patients with undiagnosed developmental disorders, and 11 events among nine patients were detected. For eight of these 11 events, mosaicism was observed in saliva but not blood, suggesting that assaying blood alone would miss a large fraction, possibly >50%, of mosaic diagnostic chromosomal rearrangements.

结构镶嵌异常是细胞子集中存在的大型合子后突变，与发育障碍和癌症有关。此类突变通常在临床诊断中使用细胞遗传学或微阵列测试进行评估。现代疾病研究严重依赖外显子组测序，但缺乏使用靶向测序数据检测结构嵌合体的适当方法。在这里，我们提出了一种名为 MrMosaic 的方法，利用等位基因分数的偏差和下一代测序数据的读取覆盖率来检测结构镶嵌异常。全外显子组测序 (WES) 和全基因组测序 (WGS) 模拟用于计算一系列镶嵌事件大小、类型、克隆性和测序深度的检测性能。该工具应用于 4911 名未确诊发育障碍患者，在 9 名患者中检测到 11 项事件。对于这 11 个事件中的 8 个，在唾液中观察到嵌合现象，但在血液中未观察到，这表明仅检测血液会漏掉很大一部分（可能>50%）的嵌合诊断染色体重排。