ABSTRACT
The use of short-read metabarcoding for classifying microeukaryotes is challenged by the lack of comprehensive 18S rRNA reference databases. While recent advances in high-throughput long-read sequencing provide the potential to greatly increase the phylogenetic coverage of these databases, the performance of different sequencing technologies and subsequent bioinformatics processing remain to be evaluated, primarily because of the absence of well-defined eukaryotic mock communities. To address this challenge, we created a eukaryotic rRNA operon clone-library and turned it into a precisely defined synthetic eukaryotic mock community. This mock community was then used to evaluate the performance of three long-read sequencing strategies (PacBio circular consensus sequencing and two Nanopore approaches using unique molecular identifiers) and three tools for resolving amplicons sequence variants (ASVs) (USEARCH, VSEARCH, and DADA2). We investigated the sensitivity of the sequencing techniques based on the number of detected mock taxa, and the accuracy of the different ASV-calling tools with a specific focus on the presence of chimera among the final rRNA operon ASVs. Based on our findings, we provide recommendations and best practice protocols for how to cost-effectively obtain essentially error-free rRNA operons in high-throughput. An agricultural soil sample was used to demonstrate that the sequencing and bioinformatic results from the mock community also translates to highly diverse natural samples, which enables us to identify previously undescribed microeukaryotic lineages.
ABSTRACT
Intensive agriculture systems have paved the way for a growing human population. However, the abundant use of mineral fertilizers and pesticides may negatively impact nutrient cycles and biodiversity. One potential alternative is to harness beneficial relationships between plants and plant-associated rhizobacteria to increase nutrient-use efficiency and provide pathogen resistance. Plant-associated microbiota profiling can be achieved using high-throughput 16S rRNA gene amplicon sequencing. However, interrogation of these data is limited by confident taxonomic classifications at high taxonomic resolution (genus- or species level) with the commonly applied universal reference databases. High-throughput full-length 16S rRNA gene sequencing combined with automated taxonomy assignment (AutoTax) can be used to create amplicon sequence variant resolved ecosystems-specific reference databases that are superior to the traditional universal reference databases. This approach was used here to create a custom reference database for bacteria and archaea based on 987,353 full-length 16S rRNA genes from Askov and Cologne soils. We evaluated the performance of the database using short-read amplicon data and found that it resulted in the increased genus- and species-level classification compared to commonly use universal reference databases. The custom database was utilized to evaluate the ecosystem-specific primer bias and taxonomic resolution of amplicon primers targeting the V5-V7 region of the 16S rRNA gene commonly used within the plant microbiome field. Finally, we demonstrate the benefits of custom ecosystem-specific databases through the analysis of V5-V7 amplicon data to identify new plant-associated microbes for two legumes and two cereal species.
