Nilsen Tonje, Snipen Lars-Gustav, Angell Inga Leena, Keeley Nigel Brian, Majaneva Sanna, Pettersen Ragnhild, Rudi Knut
Faculty of Chemistry, Biotechnology and Food Science (KBM), Norwegian University of Life Sciences (NMBU), Chr. M. Falsensvei 18, Biotechnology Building, 1432 Ås, Norway.
Institute of Marine Research, Dept. Tromso, P.O. Box 6606, Stakkevollan, 9296 Tromsø, Norway.
ISME Commun. 2024 May 9;4(1):ycae071. doi: 10.1093/ismeco/ycae071. eCollection 2024 Jan.
The performance of sequence variant resolution analytic tools for metabarcoding has not yet been adequately benchmarked for high-diversity environmental samples. We therefore evaluated the sequence variant tools DADA2, Deblur, Swarm, and UNOISE, using high-diversity seafloor samples, resulting in comparisons of 1800 sequence variant tables. The evaluation was based on 30 sediment grab samples, for which 3 replica samples were collected. Each replica sample was extracted using 5 common DNA extraction kits, resulting in 450 DNA extracts which were 16S rRNA gene sequenced (V3-V4), using Illumina. Assessments included variation across replica samples, extraction kits, and denoising methods, in addition to applying prior knowledge about alpha diversity correlations toward the cosmopolitan marine archaeon with high diversity and the sulfide oxidizing with low diversity DADA2 displayed the highest variance between replicates (Manhattan distance 1.14), while Swarm showed the lowest variance (Manhattan distance 0.93). For the analysis based on prior biological knowledge, UNOISE displayed the highest alpha diversity (Simpson's D) correlation toward (Spearman rho = 0.85), while DADA2 showed the lowest (Spearman rho = 0.10). Deblur completely eliminated from the dataset. For , on the other hand, all the methods showed comparable results. In conclusion, our evaluations show that Swarm and UNOISE performed better than DADA2 and Deblur for high-diversity seafloor samples.
对于高多样性环境样本，元条形码测序变异解析分析工具的性能尚未得到充分的基准测试。因此，我们使用高多样性的海底样本对序列变异工具DADA2、Deblur、Swarm和UNOISE进行了评估，从而得到了1800个序列变异表的比较结果。评估基于30个沉积物抓取样本，每个样本采集了3个重复样本。每个重复样本使用5种常见的DNA提取试剂盒进行提取，共得到450份DNA提取物，使用Illumina对其16S rRNA基因（V3-V4）进行测序。评估内容包括重复样本、提取试剂盒和去噪方法之间的差异，此外还应用了关于高多样性的世界性海洋古菌和低多样性的硫化物氧化菌的α多样性相关性的先验知识。DADA2在重复样本之间显示出最高的差异（曼哈顿距离为1.14），而Swarm显示出最低的差异（曼哈顿距离为0.93）。对于基于先验生物学知识的分析，UNOISE对 的α多样性（辛普森指数D）显示出最高的相关性（斯皮尔曼相关系数ρ = 0.85），而DADA2显示出最低的相关性（斯皮尔曼相关系数ρ = 0.10）。Deblur完全从数据集中消除了 。另一方面，对于 ，所有方法都显示出可比的结果。总之，我们的评估表明，对于高多样性的海底样本，Swarm和UNOISE的性能优于DADA2和Deblur。
