Effects of error, chimera, bias, and GC content on the accuracy of amplicon sequencing

作者：Yujia Qin, Liyou Wu, Qiuting Zhang, Chongqin Wen, Joy D. Van Nostrand, Daliang Ning, Lutgarde Raskin, Ameet Pinto, Jizhong Zhou · 发表于：mSystems · 年份：2023 · DOI：10.1128/msystems.01025-23 · 被引用次数：22 · 研究领域：Genomics and Phylogenetic Studies、Gut microbiota and health、Microbial Community Ecology and Physiology

ABSTRACT Targeted amplicon sequencing is widely used in microbial ecology studies. However, sequencing artifacts and amplification biases are of great concern. To identify sources of these artifacts, a systematic analysis was performed using mock communities comprised of 16S rRNA genes from 33 bacterial strains. Our results indicated that while sequencing errors were generally isolated to low-abundance operational taxonomic units, chimeric sequences were a major source of artifacts. Singleton and doubleton sequences were primarily chimeras. Formation of chimeric sequences was significantly correlated with the GC content of the targeted sequences. Low-GC-content mock community members exhibited lower rates of chimeric sequence formation. GC content also had a large impact on sequence recovery. The quantitative capacity was notably limited, with substantial recovery variations and weak correlation between anticipated and observed strain abundances. The mock community strains with higher GC content had higher recovery rates than strains with lower GC content. Amplification bias was also observed due to the differences in primer affinity. A two-step PCR strategy reduced the number of chimeric sequences by half. In addition, comparative analyses based on the mock communities showed that several widely used sequence processing pipelines/methods, including DADA2, Deblur, UCLUST, UNOISE, and UPARSE, had different advantages and disadvantages in artifact removal and rare species detec...