Global Patterns Data

Anagha Shenoy, Erin Dahl, and Lisa Karstens, PhD

June 6, 2024

Source: vignettes/microshades-GP.Rmd

Global Patterns Data Vignette

This vignette explores the Global Patterns microbiome data available from phyloseq, which includes water samples, land samples, and human samples.

Learn more about the phyloseq package here.

Additionally, the package speedyseq is necessary to use the function prep_mdf(). The package speedyseq provides faster versions of phyloseq’s plotting and taxonomic merging functions. Alternatively, the phyloseq object can be melted and transformed by using phyloseq functions tax_glom() and/or transform_sample_counts(), and melted by using psmelt().

Load the required packages 

library(microshades)
library(phyloseq)
library(ggplot2)
library(dplyr)
library(cowplot)
library(patchwork)
library(forcats)
library(tidyverse)

# The dataset Global Patterns is a phyloseq object available from the Phyloseq package
data(GlobalPatterns)

Use the microshades functions

prep_mdf

Use prep_mdf to agglomerate and normalize the phyloseq object, and melt to a data frame. Here we specify that NA values should be removed with the remove_na parameter, which can be adjusted according to the needs of your visualization and analysis.

mdf_prep <- prep_mdf(GlobalPatterns, remove_na = TRUE)

There is an alternative to using this function if you do not have speedyseq:

mdf_prep <- GlobalPatterns %>%
        tax_glom("Genus") %>%
        phyloseq::transform_sample_counts(function(x) { x/sum(x) }) %>%
        psmelt() %>%
        filter(Abundance > 0)

Both prep_mdf and the above option will produce the same results.

However, prep_mdf uses the speedyseq package to increase the speed of tax_glom and psmelt, which may be preferable when working with large datasets.

create_color_dfs

Use create_color_dfs to generate a color object for the specified data. Then extract the objects used to plot. mdf represents the object to plot; cdf represents the coloring.

color_objs_GP <- create_color_dfs(mdf_prep,
                                  selected_groups = 
                                    c("Verrucomicrobia", "Proteobacteria", "Actinobacteria", "Bacteroidetes",
                                      "Firmicutes") , 
                                  cvd = TRUE)

# Extract
mdf_GP <- color_objs_GP$mdf
cdf_GP <- color_objs_GP$cdf

Plot with default parameters

Use mdf_GP as the object to plot and use cdf_GP to assign the correct color assignments.

plot <- plot_microshades(mdf_GP, cdf_GP)

# add customizations with ggplot
plot_1 <- plot + scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.key.size = unit(0.2, "cm"), text=element_text(size=10)) +
  theme(axis.text.x = element_text(size= 6)) 

plot_1

The plot_microshades returns a ggplot object, which allows for additional specifications for the plot to be declared. For example, this allows users to facet samples and other descriptive elements.

plot_2 <- plot + scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.key.size = unit(0.2, "cm"), text=element_text(size=10)) +
  theme(axis.text.x = element_text(size= 6)) +
  facet_wrap(~SampleType, scales = "free_x", nrow = 2) +
  theme (strip.text.x = element_text(size = 6))

plot_2

Plot with custom legend

To ensure that all elements of the custom legend are visible, adjust legend_key_size and legend_text_size. If using R Markdown, it may be helpful to adjust fig.height and fig.width to receive a plot with the appropriate dimensions.

Use plot_grid from the cowplot package to plot the custom legend with the visualization.

To follow a detailed tutorial on how to use the custom_legend function, see the custom legend vignette.

GP_legend <- custom_legend(mdf_GP, cdf_GP)

plot_diff <- plot + scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.position = "none")  +
  theme(axis.text.x = element_text(size= 6)) +
  facet_wrap(~SampleType, scales = "free_x", nrow = 2) +
  theme(axis.text.x = element_text(size= 6)) + 
  theme(plot.margin = margin(6,20,6,6))

plot_grid(plot_diff, GP_legend,  rel_widths = c(1, .25))

Plot with extended group colors

Here, we plot with extended Proteobacteria colors. Note the expansion of Proteobacteria groups in the legend.

new_groups <- extend_group(mdf_GP, cdf_GP, "Phylum", "Genus", "Proteobacteria", existing_palette = "micro_cvd_orange", new_palette = "micro_orange", n_add = 5)
## Joining with `by = join_by(Top_Phylum, Top_Genus, group, hex)`
## Joining with `by = join_by(Top_Phylum, Top_Genus, group, hex)`
GP_legend_new <- custom_legend(new_groups$mdf, new_groups$cdf)

plot_diff <- plot_microshades(new_groups$mdf, new_groups$cdf) + 
  scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.position = "none")  +
  theme(axis.text.x = element_text(size= 6)) +
  facet_wrap(~SampleType, scales = "free_x", nrow = 2) +
  theme(axis.text.x = element_text(size= 6)) + 
  theme(plot.margin = margin(6,20,6,6))

plot_grid(plot_diff, GP_legend_new,  rel_widths = c(1, .25))

Plot subsets of data

Re-examine data with smaller groups by plotting subsets of the data. Here, we separate by sample type. Then, follow the prep → create → extract → plot sequence with each subset.

Water samples 

ps_water <- subset_samples(GlobalPatterns, SampleType %in% c("Freshwater", "Freshwater (creek)", "Ocean"))

mdf_water <- prep_mdf(ps_water)

color_objs_water <- create_color_dfs(mdf_water,selected_groups = c("Verrucomicrobia", "Proteobacteria", "Actinobacteria", "Bacteroidetes",
    "Firmicutes") , cvd = TRUE)

color_objs_water <- reorder_samples_by(color_objs_water$mdf, color_objs_water$cdf)

mdf_water <- color_objs_water$mdf
cdf_water <- color_objs_water$cdf
water_legend <-custom_legend(mdf_water, cdf_water)

water_plot <- plot_microshades(mdf_water, cdf_water) + 
  scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.position = "none") +
  theme(axis.text.x = element_text(size= 8)) +
  facet_wrap(~SampleType, scales = "free_x") +
  theme (strip.text.x = element_text(size = 8)) 


plot_grid(water_plot, water_legend,  rel_widths = c(1, .25))

Plot contributions of water sample types

Use plot_contributions to create median and mean abundance barplots and boxplots.

freshwater_contribution <- plot_contributions(mdf_water, cdf_water, "SampleType", "Freshwater")
creek_contribution <- plot_contributions(mdf_water, cdf_water, "SampleType", "Freshwater (creek)")
ocean_contribution <- plot_contributions(mdf_water, cdf_water, "SampleType", "Ocean")

freshwater_contribution$box + 
  creek_contribution$box + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank()) + 
  ocean_contribution$box + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank())

freshwater_contribution$mean + 
  creek_contribution$mean + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank()) + 
  ocean_contribution$mean + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank())

freshwater_contribution$median + 
  creek_contribution$median + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank()) +
  ocean_contribution$median + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank())

Land samples 

ps_land <- subset_samples(GlobalPatterns, SampleType %in% c("Soil", "Sediment (estuary)"))

mdf_land <- prep_mdf(ps_land)
color_objs_land <- create_color_dfs(mdf_land,selected_groups = c("Verrucomicrobia", "Proteobacteria", "Actinobacteria", "Bacteroidetes",
    "Firmicutes") , cvd = TRUE)

color_objs_land <- reorder_samples_by(color_objs_land$mdf, color_objs_land$cdf)

mdf_land <- color_objs_land$mdf
cdf_land <- color_objs_land$cdf
land_legend <-custom_legend(mdf_land, cdf_land)

land_plot <- plot_microshades(mdf_land, cdf_land) +
  scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.position = "none") +
  theme(axis.text.x = element_text(size= 8)) +
  facet_wrap(~SampleType, scales = "free_x") +
  theme (strip.text.x = element_text(size = 8))


plot_grid(land_plot, land_legend,  rel_widths = c(1, .25))

Plot contributions of land sample types 
sediment_contribution <- plot_contributions(mdf_land, cdf_land, "SampleType", "Sediment (estuary)")
soil_contribution <- plot_contributions(mdf_land, cdf_land, "SampleType", "Soil")


sediment_contribution$box +
  soil_contribution$box + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank())

Human samples 

ps_human <- subset_samples(GlobalPatterns, SampleType %in% c("Skin", "Feces", "Tongue"))

mdf_human <- prep_mdf(ps_human)

color_objs_human <- create_color_dfs(mdf_human,selected_groups = c("Verrucomicrobia", "Proteobacteria", "Actinobacteria", "Bacteroidetes",
    "Firmicutes") , cvd = TRUE)

color_objs_human <- reorder_samples_by(color_objs_human$mdf, color_objs_human$cdf)

mdf_human <- color_objs_human$mdf
cdf_human <- color_objs_human$cdf
human_legend <-custom_legend(mdf_human, cdf_human)

human_plot <- plot_microshades(mdf_human, cdf_human) +
  scale_y_continuous(labels = scales::percent, expand = expansion(0)) +
  theme(legend.position = "none") +
  theme(axis.text.x = element_text(size= 8)) +
  facet_wrap(~SampleType, scales = "free_x") +
  theme (strip.text.x = element_text(size = 8))


plot_grid(human_plot, human_legend,  rel_widths = c(1, .25))

Plot contributions of human sample types 
feces_contribution <- plot_contributions(mdf_human, cdf_human, "SampleType", "Feces")
skin_contribution <- plot_contributions(mdf_human, cdf_human, "SampleType", "Skin")
tongue_contribution <- plot_contributions(mdf_human, cdf_human, "SampleType", "Tongue")

feces_contribution$box +
  skin_contribution$box + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank()) +
  tongue_contribution$box + theme(axis.title.y=element_blank(), axis.text.y= element_blank(), axis.ticks.y=element_blank())