microViz: an R package for microbiome data

visualization and statistics

David J.M. Barnett1, 2 , Ilja C.W. Arts1 , and John Penders2, 3, 4
1 Maastricht Centre for Systems Biology (MaCSBio) 2 Maastricht University Medical Center+,
Department of Medical Microbiology 3 Maastricht University, Public Health Research Institute
(CAPHRI) 4 Maastricht University, School of Nutrition and Translational Research in Metabolism
(NUTRIM)
DOI: 10.21105/joss.03201
Software
• Review
Summary
• Repository
• Archive
microViz is an R package for the statistical analysis and visualization of microbiota data. This
package extends the functionality of popular microbial ecosystem data analysis R packages,
including phyloseq (McMurdie & Holmes, 2013), vegan (Oksanen et al., 2020) and microbiome
Editor: Lorena Pantano
(Lahti & Shetty, 2012-2019). microViz provides a selection of powerful additions to the toolbox
Reviewers:
of researchers already familiar with phyloseq and microbiome, as well as assisting researchers
• @marypiper with less R programming experience to independently explore and analyse their data and to
• @yoonjeongcha generate publication-ready figures.

Submitted: 09 April 2021 The tools offered by microViz include:

Published: 10 July 2021
License • A Shiny app (Chang et al., 2021) for interactive exploration of microbiota data within
Authors of papers retain R, pairing ordination plots with abundance bar charts
copyright and release the work • Easy to use functions for generating publication-ready ordination plots with ggplot2
under a Creative Commons (Wickham, 2016), accommodating constrained and partial ordination, bi-plots and tri-
Attribution 4.0 International plots, and automatic captioning designed to promote methodological transparency and
License (CC BY 4.0). reproducibility
• A novel visualization approach pairing ordination plots with circular bar charts (iris plots)
for comprehensive, intuitive and compact visualization of the similarity and composition
of hundreds of microbial ecosystems
• Correlation and composition heatmaps for microbiome data annotated with plots show-
ing each taxon’s prevalence and/or abundance
• A compact cladogram visualization approach for intuitive comparison of numerous
microbe-metadata associations derived from (multivariable) statistical models (taxo-
nomic association trees)

Statement of need

Modern microbiome research typically involves the use of next-generation-sequencing methods

to profile the relative abundance of hundreds of microbial taxa across tens, hundreds or
thousands of samples. Alongside increasing sample sizes, the amount of relevant metadata
collected is growing, particularly in human cohort studies. These trends all increase the size
and complexity of the resulting dataset, which makes its exploration, statistical analysis, and
presentation increasingly challenging.
Ordination methods like Principle Coordinates/Components Analyses (PCoA/PCA) are a sta-
ple method in microbiome research. The vegan R package implements the majority of distance

Barnett et al., (2021). microViz: an R package for microbiome data visualization and statistics. Journal of Open Source Software, 6(63), 3201. 1
https://doi.org/10.21105/joss.03201
 and ordination calculation methods and microViz makes available two further dissimilarity
measures: Generalized UniFrac from the GUniFrac package, and the Aitchison distance. The
former provides a balanced intermediate between unweighted and weighted UniFrac (Chen et
al., 2012), and the latter is a distance measure designed for use with compositional data, such
as sequencing read counts (Gloor et al., 2017).
The phyloseq R package provides an interface for producing ordination plots, with the ggplot2
R package. microViz streamlines the computation and presentation of ordination methods
including the constrained analyses: redundancy analysis (RDA), distance-based RDA, par-
tial RDA, and canonical correspondence analysis (CCA). microViz can generate highly cus-
tomizable ggplot2 bi-plots and tri-plots, showing labelled arrows for microbial loadings and
constraint variables when applicable. Furthermore, these figures are captioned automatically,
by default. The captions are intended to promote better reporting of ordination methods in
published research, where too often insufficient information is given to reproduce the ordina-
tion plot. To provide the automated captioning, microViz implements a simple S3 list class,
ps_extra, for provenance tracking, by storing distance matrices and ordination objects along-
side the phyloseq object they were created from, as well as relevant taxonomic aggregation
and transformation information.
Moreover, microViz provides a Shiny app interface (Chang et al., 2021) that allows the user to
interactively create and explore ordination plots directly from phyloseq objects. The Shiny app
generates code that can be copy-pasted into a script to reproduce the interactively designed
ordination plot. The user can click and drag on the interactive ordination plot to select samples
and directly examine their taxonomic compositions on a customizable stacked bar chart with
a clear colour scheme.
Alternatively, for a comprehensive and intuitive static presentation of both sample variation
patterns and underlying microbial composition, microViz provides an easy approach to pair
ordination plots with attractive circular bar charts (iris plots) by ordering the bar chart in
accordance with the rotational position of samples around the origin point on the ordination
plot, e.g. Figure 1. Bar charts do have limitations when visualizing highly diverse samples, such
as the adult gut microbiome, at a detailed taxonomic level. This is why microViz also offers an
enhanced heatmap visualization approach, pairing an ordered heatmap of (transformed and/or
scaled) microbial abundances with compact plots showing each taxon’s overall prevalence
and/or abundance distribution. The same annotation can easily be added to metadata-to-
microbe correlation heatmaps.
microViz provides a flexible wrapper around methods for the statistical modelling of microbial
abundances, including e.g. beta-binomial regression models from the corncob R package (Mar-
tin et al., 2020), and compositional linear regression. To visualize metadata-to-microbiome
associations derived from more complicated statistical models, microViz offers a visualization
approach that combines multiple annotated cladograms to comprehensively and compactly
display patterns of microbial associations with multiple covariates from the same multivari-
able statistical model. These “taxonomic association trees” facilitate direct comparison of the
direction, strength and significance of microbial associations between covariates and across
multiple taxonomic ranks. This visualisation also provides an intuitive reminder of the bal-
ancing act inherent in compositional data analysis: if one clade/branch goes up, others must
go down. Other packages in R, such as ggtree (Yu et al., 2017) or metacoder (Foster et
al., 2017), can be used to make annotated cladograms similar to the microViz taxonomic
association tree visualizations, but the microViz style has a few advantages for the purpose
of reporting multivariable model results: Firstly, microViz cladogram generating functions are
directly paired with functions to compute the statistical model results for all taxa in a phy-
loseq. Secondly, the tree layouts are more compact, by default, for displaying multiple trees
for easy comparison.
Finally, beyond the main visualization functionality, microViz provides a suite of tools for
working easily with phyloseq objects including wrapper functions that bring approaches from

Barnett et al., (2021). microViz: an R package for microbiome data visualization and statistics. Journal of Open Source Software, 6(63), 3201. 2
https://doi.org/10.21105/joss.03201
 the popular dplyr package to phyloseq, to help the researcher easily filter, select, join, mu-
tate and arrange phyloseq sample data. All microViz functions are designed to work with
magrittr’s pipe operator (%>%), to chain successive functions together and improve code
readability (Bache & Wickham, 2020). Lastly, for user convenience, microViz documentation
and tutorials are hosted online via a pkgdown (Wickham & Hesselberth, 2020) website on
Github Pages, with extensive examples of code and output generated with example datasets.

Figure 1: Simple example of a microViz figure pairing an ordination plot of microbial samples (left)
with an “iris plot” (right): a circular stacked barchart showing the microbial compositions of samples
ordered in accordance with the ordination plot. This figure is created with a subset of the “dietswap”
dataset available within the microbiome R package. The ordination plot is a PCA bi-plot created
using centered-log-ratio transformed species-like HITChip microbial features. The dark grey filled
points on both plots indicate samples where the participant’s nationality is AFR. AFR = African;
AAM = African American.

Acknowledgements

This work was completed as part of a project jointly funded by the Dutch Research Coun-
cil (NWO), AVEBE, FrieslandCampina and NuScience, as coordinated by the Carbohydrate
Competence Center (CCC-CarboBiotics; www.cccresearch.nl).

References

Bache, S. M., & Wickham, H. (2020). Magrittr: A forward-pipe operator for r. https:
//CRAN.R-project.org/package=magrittr
Chang, W., Cheng, J., Allaire, J., Sievert, C., Schloerke, B., Xie, Y., Allen, J., McPherson,
J., Dipert, A., & Borges, B. (2021). Shiny: Web application framework for r. https:
//CRAN.R-project.org/package=shiny
Chen, J., Bittinger, K., Charlson, E. S., Hoffmann, C., Lewis, J., Wu, G. D., Collman,
R. G., Bushman, F. D., & Li, H. (2012). Associating microbiome composition with
environmental covariates using generalized UniFrac distances. Bioinformatics, 28(16),
2106–2113. https://doi.org/10.1093/bioinformatics/bts342
Foster, Z., Sharpton, T., & Grünwald, N. (2017). Metacoder: An R package for visualization
and manipulation of community taxonomic diversity data. PLOS Computational Biology,
13(2), 1–15. https://doi.org/10.1371/journal.pcbi.1005404

Barnett et al., (2021). microViz: an R package for microbiome data visualization and statistics. Journal of Open Source Software, 6(63), 3201. 3
https://doi.org/10.21105/joss.03201
 Gloor, G. B., Macklaim, J. M., Pawlowsky-Glahn, V., & Egozcue, J. J. (2017). Microbiome
Datasets Are Compositional: And This Is Not Optional. Frontiers in Microbiology, 8,
2224. https://doi.org/10.3389/fmicb.2017.02224
Lahti, L., & Shetty, S. (2012-2019). Microbiome r package: Tools for microbiome analysis in
r. https://github.com/microbiome/microbiome
Martin, B. D., Witten, D., & Willis, A. D. (2020). Modeling microbial abundances and
dysbiosis with beta-binomial regression. The Annals of Applied Statistics, 14(1), 94–115.
https://doi.org/10.1214/19-AOAS1283
McMurdie, P. J., & Holmes, S. (2013). Phyloseq: An R package for reproducible interactive
analysis and graphics of microbiome census data. PLoS ONE, 8(4), e61217. https:
//doi.org/10.1371/journal.pone.0061217
Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P.
R., O’Hara, R. B., Simpson, G. L., Solymos, P., Stevens, M. H. H., Szoecs, E., & Wagner,
H. (2020). Vegan: Community Ecology Package. https://CRAN.R-project.org/package=
vegan
Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New
York. ISBN: 978-3-319-24277-4
Wickham, H., & Hesselberth, J. (2020). Pkgdown: Make static HTML documentation for a
package. https://CRAN.R-project.org/package=pkgdown
Yu, G., Smith, D. K., Zhu, H., Guan, Y., & Lam, T. T.-Y. (2017). Ggtree: An r package
for visualization and annotation of phylogenetic trees with their covariates and other as-
sociated data. Methods in Ecology and Evolution, 8, 28–36. https://doi.org/10.1111/
2041-210X.12628

Barnett et al., (2021). microViz: an R package for microbiome data visualization and statistics. Journal of Open Source Software, 6(63), 3201. 4
https://doi.org/10.21105/joss.03201