Package 'plotly.microbiome'

Draw 3D Scatterplot of Beta Diversity

Description

This is a wrapper function for plot_ly with the "scatterplot3d" option. It can be used on a data frame, or directly on the output of phyloseq::ordinate with the NMDS or PCoA method (or vegan::metaMDS or ape::pcoa, respectively), or on the output of limma::plotMDS. Default categorical colors are from dittoSeq.

Usage

beta_diversity_3d(x, ...)

## S3 method for class 'data.frame'
beta_diversity_3d(x, axes = colnames(x)[1:3],
                  color.column = colnames(x)[4],
                  label.column = colnames(x)[5],
                  color.key = NULL, ...)

## S3 method for class 'monoMDS'
beta_diversity_3d(x, metadata,
                  color.column,
                  label.column = NULL,
                  color.key = NULL, ...)

## S3 method for class 'pcoa'
beta_diversity_3d(x, metadata,
                  color.column,
                  label.column = NULL,
                  color.key = NULL, ...)

## S3 method for class 'MDS'
beta_diversity_3d(x, metadata,
                  color.column,
                  label.column = NULL,
                  color.key = NULL, ...)

Arguments

x	A data frame, "monoMDS", "pcoa", or "MDS" object containing ordination results.
axes	If x is a data frame, the column names for three axes to plot.
color.column	The name of a numeric or categorical column to be used for coloring points. This column should be found in x if x is a data frame, or in metadata otherwise.
label.column	The name of a character column to be used for labeling points. This column should be found in x if x is a data frame, or in metadata otherwise. It defaults to sample names if x is a "monoMDS" or "MDS".
color.key	If color.column refers to a character or factor column, a named vector of colors, with names corresponding to values in the column. The default is to use dittoSeq::dittoColors. If color.column refers to a numeric column, a long vector of colors to be used as the color scale. The default is to use viridis. Passed to the colors argument of plot_ly.
metadata	A data frame of sample metadata, in the same order as x.
...	Optional arguments passed to plot_ly.

Value

A "plotly" object.

Author(s)

Lindsay Clark

Examples

## Not run: 
# Perform NMDS with Bray distance

ord_mfiber_prop1a <- ordinate(ps_mfiber_prop, "NMDS", "bray", k = 3)

beta_diversity_3d(ord_mfiber_prop1a,
                  sample_data(ps_mfiber_prop), "TRT", "Label")

# NMDS with UniFrac distance (ordinate won't let you adjust k)

dist_mfiber_4 <- phyloseq::distance(ps_mfiber_prop, "unifrac")
ord_mfiber_prop4a <- vegan::metaMDS(dist_mfiber_4, k = 3)

beta_diversity_3d(ord_mfiber_prop4a,
                  sample_data(ps_mfiber_prop), "TRT", "Label")
                  
# MDS of gene expression with limma
mds1 <- plotMDS(logCPM.filt, top = 5000)
beta_diversity_3d(mds1, metadata = d.filt$samples,
                  color.column = "Group")

# Changing title and axis labels, manually setting colors, and saving

mfiber_colors <- c(CO = "magenta", BP = "black", MF = "turquoise",
                   FOS = "orange", RS = "skyblue", TP = "green")

p1 <- beta_diversity_3d(ord_mfiber_prop1a,
                  sample_data(ps_mfiber_prop), "TRT", "Label",
                  color.key = mfiber_colors) %>%
  plotly::layout(title = "NMDS with Bray distance",
         scene = list(xaxis = list(title = "Axis 1"),
                      yaxis = list(title = "Axis 2"),
                      zaxis = list(title = "Axis 3")))
htmlwidgets::saveWidget(partial_bundle(p1),
                        file = "results/NMDS_Bray_MFiber_plotly.html")

## End(Not run)

---

Prepare a Data Frame for Composition Plots

Description

This function takes a phyloseq object and prepares a data frame that can be used to generate composition plots with ggplot2. Works similarly to the ggformat function in phyloseq.extended.

Usage

composition_df(psobj, rank = "Family",
               keepcols = c("Sample", "Group", "Label", "ID"),
               minprop = 0.05, mean_across_samples = NULL)

Arguments

psobj	A phyloseq object containing raw, potentially agglomerated, counts.
rank	The taxonomic rank across which taxa counts shoud be summed.
keepcols	Names of columns from sample_data(psobj)
minprop	Threshold for showing a taxon vs. lumping it into "Other". At least one sample must have at least this proportion of the OTU counts assigned to a given taxon for that taxon to be displayed.
mean_across_samples	An optional grouping variable, indicated as a character string matching one of the column names in keepcols. If provided, samples are lumped within groups. Proportions are averaged across samples, and counts are summed across samples.

Value

A tibble with the following columns:

• A column labeled ID listing sample names, or if provided, a column named the same as mean_across_samples.

• A column named the same as rank listing taxa names.

• A column named Proportion indicating the proportion of OTU counts assigned to a given taxon within a sample or group.

• A column named Counts indicating the total OTU counts assigned to a given taxon within a sample or group.

• Any other columns in keepcols. If mean_across_samples is provided, columns are dropped if they have more than one value for a given group.

Author(s)

Lindsay Clark

See Also

findnonmissing is used to determine which taxa to label as “Unclassified”.

Examples

## Not run: 
# Columns to keep
kc <- c("Dog", "Trt", "Day", "Breed", "Group", "Label", "ID")

# Composition plot on individuals, grouped by experimental group
p1 <- composition_df(ps_glom, "Family", minprop = 0.1,
               keepcols = kc) 
  ggplot(aes(x = ID, y = Proportion, fill = Family)) +
  geom_col() +
  facet_wrap(~ Group, scales = "free_x") +
  scale_fill_manual(values = dittoSeq::dittoColors(1)) +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))

ggplotly(p1)

# Composition plot on treatments
p2 <- composition_df(ps_glom, "Family", minprop = 0.1,
               keepcols = kc, mean_across_samples = "Trt") 
  ggplot(aes(x = Trt, y = Proportion, fill = Family)) +
  geom_col() +
  scale_fill_manual(values = dittoSeq::dittoColors(1)) +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))

ggplotly(p2)

## End(Not run)

---

Identify Non-Missing Taxonomic Labels

Description

In various taxonomic databases and pipelines, unknown or missing taxonomic labels may be indicated in a variety of ways, such as missing data, "unclassified", "uncultured", etc. This function identifies all of these that I have encountered so far.

Usage

findnonmissing(x)

Arguments

x	A character vector of taxonomic labels, for example a single column of the tax_table slot of a phyloseq object.

Details

The following values will result in output of FALSE.

• NA

• An empty string.

• The words “unclassified”, “unidentified”, “uncultured”, “unknown”, or “metagenome” anywhere in the string, in any case.

• Values equal to “human_gut.”

Value

A logical vector, with TRUE if the taxonomic label reflects a taxonomic identity, and FALSE if it should be considered missing.

Author(s)

Lindsay V. Clark

Examples

findnonmissing(c("Streptococcus", "Blautia", "Horse metagenome", NA))

---

Create a Label for Each Taxon

Description

This function takes a table of taxonomic ranks, such as that stored in the tax_table slot of a phyloseq object, and creates a label for each taxon for use in plots and tables.

Usage

make_taxa_labels(taxtab)

Arguments

taxtab

A matrix or data frame, with taxa in rows and taxonoic ranks in columns. The last column should be “Species”, the first column should be kingdom or domain, and columns in between should progress in order of rank.

Details

Species labels that pass findnonmissing are used, and otherwise species are labeled “"sp."”. The lowest rank that passes findnonmissing is pasted before the species label.

Value

A character vector containing the labels. If taxtab has row names, these are used to name the vector.

Author(s)

Lindsay V. Clark

Examples

tt <- matrix(c("Bacteria", "Firmicutes", "Clostridia",
               "Peptostreptococcales-Tissierellales", "Anaerovoracaceae",
               "Mogibacterium", "Unclassified",
               "Bacteria", "Firmicutes", "Clostridia", "Oscillospirales",
               "Ruminococcaceae", "Faecalibacterium", "prausnitzii",
               "Bacteria", "Firmicutes", "Clostridia",
               "Clostridia vadinBB60 group", "Unclassified", "Unclassified",
               "Unclassified"),
               nrow = 3, ncol = 7, byrow = TRUE,
               dimnames = list(c("db3b201de3a824d7356ce4d8360e5bc3",
                                 "615ce01e68e098adc445d06e072ba255",
                                 "6f643e7faeab3e737b38f5304b841d97"),
                               c("Kingdom", "Phylum", "Class", "Order",
                                 "Family", "Genus", "Species")))

make_taxa_labels(tt)

---