Bioconductor Workflow for Microbiome Data Analysis: from raw reads to community analyses

Abstract

High-throughput sequencing of PCR-amplified taxonomic markers (like the 16S rRNA gene) has enabled a new level of analysis of complex bacterial communities known as microbiomes. Many tools exist to quantify and compare abundance levels or OTU composition of communities in different conditions. The sequencing reads have to be denoised and assigned to the closest taxa from a reference database. Common approaches use a notion of 97% similarity and normalize the data by subsampling to equalize library sizes. In this paper, we show that statistical models allow more accurate abundance estimates. By providing a complete workflow in R, we enable the user to do sophisticated downstream statistical analyses, whether parametric or nonparametric. We provide examples of using the R packages dada2, phyloseq, DESeq2, ggplot2 and vegan to filter, visualize and test microbiome data. We also provide examples of supervised analyses using random forests and nonparametric testing using community networks and the ggnetwork package. Keywords Amendments from Version 1In version 2 of the manuscript:We have updated the procedure for storing the filtered and trimmed files during the call to dada2, this avoids overwriting the files if the workflow is run several times.We have replaced the msa alignment function with AlignSeqs function from from the DECIPHER 1 package, making the workflow more computationally efficient.We have expanded the phyloseq section and reduced the number of network plots. We have also provided detailed discussion of our choice not to make the PCoA and PCA plots square.We have added more detailed instructions in the Github repository as to how one can run only parts of the workflow and how to generate the full paper from scratch using the main.rnw file.As suggested by reviewers, we have added more extended captions to figures. We have however refrained from providing a complete evaluation of DADA2 vs. OTUs or pooled/unpooled data to this manuscript. Performing such evaluations well is a significant undertaking and would take significant space to explain, and our primary purpose here is to demonstrate the many features of an R/Bioconductor amplicon analysis workflow.We thank the three reviewers and a commentator who have provided useful feedback and we hope the revision has enhanced the readability and explained the code more completely.

Testing clustering strategies for metabarcoding‐based investigation of community–environment interactions

“…Recently, nonclustering strategies (e.g., DADA2 and UNOISE3) were held in esteem in numerous studies to generate ZOTUs (Hugerth & Andersson, ), as these methods improved resolution on biodiversity assessment, which was a cornerstone for ecological studies (Callahan et al., ; Edgar, ). Interestingly, we recapitulated the results of nonclustering methods when we explored community–environment relationship by the use of clustering thresholds of 99%–97% in the present study.…”

“…In the DADA2 workflow, sequence reads were filtered using the standard filtering parameters (i.e., maxN = 0, truncQ = 2, rm.phix = TRUE and maxEE = 2) in the online pipeline suggested by Callahan et al. () and trimmed to 225 bp as suggested by our previous study (Xiong et al., ). Then error model parameter learning was carried out on the sequences passing the filtering step.…”

“…After robust denoizing processes, these new methods generate ZOUTs by producing unique sequences in a data set (Edgar, ). A ZOUT is represented by a unique sequence in a data set, thus eliminating the lumping process during sequence data processing (Callahan et al., ; Edgar, ). As taxonomically different organisms with distinct sequences may be merged into one single OTU in traditionally used clustering strategies, ZOTUs have improved resolution of metabarcoding by exactly resolving amplicon sequence variant down to the level of single nucleotide difference (Callahan et al., ).…”

“…A ZOUT is represented by a unique sequence in a data set, thus eliminating the lumping process during sequence data processing (Callahan et al., ; Edgar, ). As taxonomically different organisms with distinct sequences may be merged into one single OTU in traditionally used clustering strategies, ZOTUs have improved resolution of metabarcoding by exactly resolving amplicon sequence variant down to the level of single nucleotide difference (Callahan et al., ). Several studies have approved that ZOTUs are superior to conventional OTUs for most study purposes by the improved resolution (e.g., Callahan, DiGiulio, et al., ; Edgar, ).…”

No differences in microbiome changes between anti‐adhesive and antibacterial ingredients in toothpastes during periodontal therapy

“…Trimmed (raw) reads were then processed using the R language environment v.3.5.0 and RStudio v.1.1.447, following the DADA2 v.1.8.0 workflow described by Callahan et al . Applied pipeline settings were explained in detail, before .…”