MetaWorks: A flexible, scalable bioinformatic pipeline for high-throughput multi-marker biodiversity assessments

Operational taxonomic units This pipeline supports the analysis of ESVs for the additional genetic and taxonomic resolution provided by this level of analysis [24]. Though this method of analysis was initially used to pro- cess 16S rRNA genes, studies using ITS and COI have also shown that the analysis of ESVs improves the detection of genetic diversity and richness, when assessing beta diversity, both ESVs and OTUs tend to recover similar gradients in multivariate analyses [25,26]. Although it has been shown that for many clustering methods sequence order matters and OTU composi- tion can change from one analysis to the next making reproducibility an issue, there are several reasons why a user would still want to analyze OTUs. For example, it may be more advanta- geous to work with OTUs instead of ESVs for network analysis to detect more co-occurrences, for legacy reasons to compare results to previous studies that used OTUs, or to approximate ‘species’ units [48]. After processing raw reads using the snakefile_ESV workflow described in the previous sec- tion, users can use the snakefile_OTU workflow to cluster ESVs into OTUs. This approach combines the benefits of denoising with clustering using a 97% sequence similarity cutoff using the snakefile_OTU workflow [26,49]. This method uses VSEARCH ‘cluster_smallmem’ method to cluster ESVs using a 97% sequence similarity cutoff. Settings can be adjusted in the in the config_OTU.yaml file such as pointing to the directory that contains the ESVs and choosing a classifier for the OTUs. Results and discussion

MetaWorks has already been used in several publications for the Canadian STREAM biomoni- toring program, the Government of Canada, Genomics Research and Development Initiative, Metagenomics-based ecosystem biomonitoring (Ecobiomics) project, and by Natural Resources Canada [18,50,51]. The benefits of using an automated, scalable, versioned pipeline for biomonitoring are many-fold, from the ability to share reproducible workflows with collab- orators to facilitate the re-analysis of data as more samples are collected from year-to-year. We describe three MetaWorks use-cases in more detail below. Use case 1: As a part of the Canadian STREAM biomonitoring initiative, the MetaWorks pipeline has been used to process macroinvertebrate COI metabarcodes surveyed from stream sites across Canada [50]. One feature of this project is the quick 1–2 month turn-around time from sampling through to the production of watershed biodiversity reports. This is an improvement over reports generated using conventional morphology-based methods that would normally take 6–12 months to produce. The use of a consistent bioinformatics work- flow to process metabarcodes has played a key role in the reproducibility, scalability, and throughput to facilitate timely reporting [52]. Generally, samples are processed in batches of 96 per sequencing run then later split into custom reports for stakeholders, processing about 500 samples per year. One feature of these reports are the taxonomic assignments made using the naive Bayesian classifier that provides bootstrap support values. During the data analysis

stage, users can use minimum bootstrap support cutoffs to ensure a certain level of expected accuracy (80–99%) and reduce false-positive taxonomic assignments [28]. The cutoffs used are specific to the amplicon, amplicon length, and taxonomic rank of the assignment and assumes the query is represented in the underlying sequence database. This is in contrast with the use of more traditional methods for taxonomic assignment, where taxa are routinely missed dur- ing subsampling and taxa detected by primary analysts and auditors may differ by up to 30% [53]. This use-case shows how MetaWorks can be used to create taxon lists for large-scale bio- diversity monitoring of streams across Canada. PLOS ONE MetaWorks mult-marker bioinformatic pipeline PLOS ONE | https://doi.org/10.1371/journal.pone.0274260 September 29, 2022 6 / 11