Thermus thermophilus DNA can be used as internal control for process monitoring of clinical metagenomic next-generation sequencing of urine samples

IntroductionProcess control for clinical metagenomic next-generation sequencing (mNGS) is not yet widely applied, while technical sources of bias are plentiful. If mNGS is to be safely used for clinical microbiology diagnostics, potential sources of error or variation in library preparation and sequencing need to be controlled for. AimWe aimed to develop an easy-to-use internal control (IC) method focussing on technical process control of library preparation and sequencing applied to mNGS in clinical microbiology diagnostics.MethodologyDNA of nine urine samples was sequenced on the Ion Torrent Proton and PGM in the absence and presence of Thermus thermophilus DNA as IC in incremental concentrations (0.5-2-5%). Between aliquots of each sample, we compared the IC relative abundance (RA), and after in silico subtraction of IC reads, the microbiota and the RA of pathogens. The proportion of cross-contaminant IC reads was determined by sequencing a replicate the absence of IC. The optimal IC spike-in concentration was defined as the lowest concentration still detectable in all samples with the smallest impact on the microbial community.ResultsThe RA of IC correlated linearly with the spiked IC concentration (r2=0.99). IC added in a concentration of 0.5% of total DNA concentration was detectable in all samples, regardless of human/bacterial composition and after in silico removal gave the smallest difference in RA of pathogens compared to the unspiked aliquot of the sample. Cross-contamination of IC reads was 0.001%. The microbiota of sample aliquots sequenced in the presence and absence of IC was highly similar after in silico removal of IC reads (median BC-dissimilarity per sample of 0.059), provided samples had sufficient bacterial read counts. ConclusionT. thermophilus DNA at a percentage of 0.5% of the total DNA concentration was successfully applied for the process control of mNGS of urine samples. We demonstrated negligible alterations in sample microbial composition after in silico subtraction of IC sequence reads. This approach contributes toward implementation of mNGS in the clinical microbiology laboratory.

引言 临床宏基因组下一代测序（metagenomic next-generation sequencing, mNGS）的过程控制尚未得到广泛应用，且存在大量技术偏倚来源。若要将mNGS安全应用于临床微生物学诊断，需对文库制备与测序环节中的潜在误差或变异来源进行管控。
研究目的 本研究旨在开发一种易于操作的内参（internal control, IC）方法，聚焦于临床微生物学诊断中用于mNGS的文库制备与测序技术过程控制。
研究方法 我们将9份尿液样本的DNA，在分别添加梯度浓度（0.5%、2%、5%递增）的嗜热栖热菌（Thermus thermophilus）DNA作为内参的情况下，于Ion Torrent Proton和PGM测序平台上完成测序。针对每份样本的不同分装样本，我们比较了内参的相对丰度（relative abundance, RA）；并在通过虚拟计算去除内参读数后，对比了样本的微生物组构成以及病原体的相对丰度。通过在不添加内参的条件下对样本进行重复测序，我们测定了交叉污染的内参读数占比。最优内参添加浓度被定义为：可在所有样本中被检出的最低浓度，且对微生物群落的影响最小。
研究结果 内参的相对丰度与添加的内参浓度呈线性相关（r²=0.99）。当添加占总DNA浓度0.5%的内参时，无论样本的人类/细菌组成如何，所有样本均可检出该内参；且经虚拟去除内参读数后，与未添加内参的样本分装相比，病原体相对丰度的差异最小。内参读数的交叉污染率为0.001%。在虚拟去除内参读数后，添加与未添加内参的样本分装的微生物组构成高度相似（每份样本的布莱-柯蒂斯相异度（Bray-Curtis dissimilarity）中位数为0.059），前提是样本具备足够的细菌读数数量。
研究结论 我们成功将占总DNA浓度0.5%的嗜热栖热菌DNA应用于尿液样本mNGS的过程控制。研究证实，经虚拟去除内参序列读数后，样本微生物组构成的改变可忽略不计。该方法有助于推动mNGS在临床微生物实验室中的落地应用。