法医学二代测序STR分型准确度与测序深度的关联性评估

Correlativity between Sequencing Depth of Next Generation Sequencing and Its Resulting Accuracy for Forensic STR Genotyping

目的利用实验数据对法医学二代测序STR分型测序深度与分型结果准确度的关联性进行评估。方法使用商业化基因组DNA制备单一来源和混合的DNA样本,以Thermo Fisher公司的25重早期测试试剂盒进行目的STR片段扩增,每种扩增产物分别使用4种不同的序列标签平行建库,并控制标记每一种序列标签的文库上机量依次占一张Ion 318芯片的1/4、1/8、1/16、1/32。经Ion PGMTM基因测序仪测序,以及Ion Torrent SuiteTM软件进行数据分析;同时对庞敬博等人发表的基于相同试剂盒和测序仪检测的95名中国汉族无关个体的6928条等位基因、影子峰和噪音序列进行测序深度统计分析,寻找测序深度与STR分型准确度的关联性。结果各基因座测序深度随文库上样量减少而呈明显下降趋势。对于单一来源样本,每张芯片上样不超过8个均一化文库可实现全部基因座的完整分型;对于1∶20比例的混合DNA,每张芯片上样不超过4个均一化文库时,未发现微量组分的等位基因丢失。人群数据测序深度统计显示,该体系基因座间存在不均衡性,有必要针对各基因座分别设定分析阈值参数。结论测序深度与法医学STR分型结果的准确性密切相关,各基因座最低测序深度与平均测序深度的比值可作为设定分析阈值的重要参考指标。本研究确定的单张芯片上样数量仅适用于本实验体系,但相关实验设计和方案可供其他实验体系开展类似工作参考。

Objective To assess the correlation between sequencing depth of next generation sequencing(NGS) and its resulting accuracy for forensic STR genotyping. Methods Commercial products of genomic DNA were selected to prepare single-sourced and mixed DNA samples. The targeted STR-fragment amplification was carried out with a kit of 25-plex early access STR panel from Thermo Fisher Scientific, therewith having four libraries resulted through each amplification product and different barcode adapters. The amount of libraries, linked with each barcode adapter, was controlled to respectively occupy 1/4, 1/8, 1/16 and 1/32 of one Ion 318 chip. The pooled libraries were sequenced on an Ion PGMTM machine, with the sequencing data being analyzed by way of the Ion Torrent SuiteTM software. Meanwhile, deep exploration was conducted into a dataset of 6928 sequences representing alleles, stutters and noises that were harvested from 95 unrelated Chinese Hanethnic individuals who were the subjects of one project accomplished and reported by Pang Jingbo et al using the same kits and sequencer as this work. Thus, the NGS sequencing depth was correlatively probed with the accuracy of STR genotyping. Results The sequencing depths of each STR locus decreased significantly with the declining amounts of loaded libraries. For single-sourced samples, full genotyping profiles can be obtained when no more than 8 normalized libraries were loaded to a single chip. For mixed DNA samples at a ratio of 1 ∶ 20, no dropout allele was observed from the minor contributor when no more than 4 normalized libraries were sequenced on one chip. Sequencing depth statistics showed that the coamplification system was not balanced among STR loci, suggesting the necessity of setting an analysis threshold for each locus. Conclusions Sequencing depth closely correlates with the accuracy of forensic STR genotyping. For each locus, the ratio of minimum sequencing depth to the average can be an important indicator for setting an analysis threshold. The number of libraries loaded to a single chip depends on the available kits, machines and experimental procedures with all of which are yet applicable for the same or similar task to get reference and suggestions from this work.