背景:用Illumina技术对16SrRNA基因(300bp)的可变区进行测序通常用于研究人类微生物群的组成。不幸的是,短读段无法区分高度相似的物种。考虑到来自同一属的物种可能与健康或疾病相关,重要的是以最低的分类学等级鉴定它们。第三代测序平台,如PacBioSMRT,增加读段长度,允许以最大的分类学分辨率对整个基因进行测序。尽管有潜力,全长16SrRNA基因测序尚未广泛使用。当前研究的目的是比较两种方法(Illumina短读测序和PacBio长读测序16SrRNA基因)在不同人类微生物组样品中的测序输出和分类学注释性能。唾液中的DNA,分离了9名志愿者的口腔生物膜(龈下菌斑)和粪便。通过IlluminaMiseq和PacBioSequelII测序仪扩增和测序区域V3-V4和V1-V9,分别。
结果:在这两个平台上,相似百分比的读数被分配到属水平(分别为94.79%和95.06%),但对于PacBio,更高比例的读数被进一步分配到物种水平(55.23%对74.14%).关于总体细菌组成,样品通过小生境而不是通过测序平台进行聚类。此外,对于所有类型的样品,在两个平台中均检测到丰度>0.1%的所有属.尽管某些属如链球菌倾向于在PacBio中观察到的频率高于在Illumina中观察到的频率(在唾液中为20.14%vs14.12%,10.63%和6.59%在龈下菌斑生物膜样品中)校正多次测试时,差异均无统计学意义。
结论:当前手稿中呈现的结果表明,使用Illumina和PacBio测序的样品大多具有可比性。考虑到PacBio读数比Illumina更准确地在物种水平上分配,我们的数据支持将PacBio技术用于未来的微生物组研究,尽管目前需要更高的成本来获得每个样品的等量读数。
BACKGROUND: Sequencing variable regions of the 16S rRNA gene (≃300 bp) with Illumina technology is commonly used to study the composition of human microbiota. Unfortunately, short reads are unable to differentiate between highly similar species. Considering that species from the same genus can be associated with health or disease it is important to identify them at the lowest possible taxonomic rank. Third-generation sequencing platforms such as PacBio SMRT, increase read lengths allowing to sequence the whole gene with the maximum taxonomic resolution. Despite its potential, full length 16S rRNA gene sequencing is not widely used yet. The aim of the current study was to compare the sequencing output and taxonomic annotation performance of the two approaches (Illumina short read sequencing and PacBio long read sequencing of 16S rRNA gene) in different human microbiome samples. DNA from saliva, oral biofilms (subgingival plaque) and faeces of 9 volunteers was isolated. Regions V3-V4 and V1-V9 were amplified and sequenced by Illumina Miseq and by PacBio Sequel II sequencers, respectively.
RESULTS: With both platforms, a similar percentage of reads was assigned to the genus level (94.79% and 95.06% respectively) but with PacBio a higher proportion of reads were further assigned to the species level (55.23% vs 74.14%). Regarding overall bacterial composition, samples clustered by niche and not by sequencing platform. In addition, all genera with > 0.1% abundance were detected in both platforms for all types of samples. Although some genera such as Streptococcus tended to be observed at higher frequency in PacBio than in Illumina (20.14% vs 14.12% in saliva, 10.63% vs 6.59% in subgingival plaque biofilm samples) none of the differences were statistically significant when correcting for multiple testing.
CONCLUSIONS: The results presented in the current manuscript suggest that samples sequenced using Illumina and PacBio are mostly comparable. Considering that PacBio reads were assigned at the species level with higher accuracy than Illumina, our data support the use of PacBio technology for future microbiome studies, although a higher cost is currently required to obtain an equivalent number of reads per sample.