Can I do sampleAnnotation() to comp_barplot object as in comp_heatmap?

[Stray357]

I would really like to annotate my sample group in my comp_barplot() object as colored bars below the axis, just like the online tutorial for comp_heatmap.( https://david-barnett.github.io/microViz/ just above "Example ordination plot workflow" section, the annotation part circled in red below)

So I tried directly adding + samplrAnnotation() to my comp_barplot object and got an error as 'Can't add sampleAnnotation(treatment_group = anno_sample_cat(var = "group", to a ggplot object.'

I guess sampleAnnotation() just doesn't work that way.

But please, is there any way to do the annotation the way it can be easily done for comp_heatmap? I've noticed that the default order of the sample in comp_barplot is set by their “bray”-curtis dissimilarity, so it would help a lot if information about experiment group can be annotated on the plot at the same time.
Many, many thanks!

[david-barnett]

here is a somewhat hacky way to do this for one sample variable

library(microViz)
library(ggplot2)
library(dplyr)
data(dietswap, package = "microbiome")

# how to add coloured segments to the left of (or underneath) stacked bars
dietswap %>%
  ps_filter(timepoint == 1) %>% 
  comp_barplot(
    tax_level = "Genus", n_taxa = 15, other_name = "Other",
    taxon_renamer = function(x) stringr::str_remove(x, " [ae]t rel."),
    palette = distinct_palette(n = 15, add = "grey90"),
    merge_other = FALSE, bar_outline_colour = "darkgrey"
  ) +
  geom_segment(
    mapping = aes(colour = nationality, xend = SAMPLE), # colour, but not fill
    data = ~ dplyr::distinct(., SAMPLE, .keep_all = TRUE), # otherwise many segments are (over)plotted per sample, which looks rough
    y = -.075, yend = -0.025, # set position of segments UNDER "axis"
    size = 3.5 # thickness of bar, manually adjust this for your plot
  ) +
  scale_y_continuous(
    limits = c(-0.05, 1.01), # adds space in the plot panel for the extra squares
    breaks = c(0, .25, .5, .75, 1) # redraws original axis breaks
  ) +
  scale_colour_manual(values = c("grey15", "grey65")) +
  guides(
    # remove black squares from the fill legend
    fill = guide_legend(reverse = TRUE, override.aes = list(colour = "transparent"))
  ) +
  theme(legend.key = element_blank()) + # remove grey background from colour legend
  coord_flip()

# an alternative with coloured squares above / to the right of bars

dietswap %>%
  ps_filter(timepoint == 1) %>% 
  comp_barplot(
    tax_level = "Genus", n_taxa = 15, other_name = "Other",
    taxon_renamer = function(x) stringr::str_remove(x, " [ae]t rel."),
    palette = distinct_palette(n = 15, add = "grey90"),
    merge_other = FALSE, bar_outline_colour = "darkgrey"
  ) +
  geom_point(
    mapping = aes(colour = nationality), # colour, but not fill
    data = ~ dplyr::distinct(., SAMPLE, .keep_all = TRUE), # otherwise many squares are (over)plotted per sample, which looks rough
    y = 1.05, # set position of squares
    size = 2.5,
    shape = "square" # must be a solid shape, as only colour (but not fill) mapping is possible
  ) +
  scale_y_continuous(
    limits = c(0, 1.2), # adds space in the plot panel for the extra squares
    breaks = c(0, .25, .5, .75, 1) # redraws original axis breaks
  ) +
  scale_colour_manual(values = c("grey15", "grey65")) +
  guides(
    # remove black squares from the fill legend
    fill = guide_legend(reverse = TRUE, override.aes = list(colour = "transparent"))
  ) +
  theme(legend.key = element_blank()) + # remove grey background from colour legend
  coord_flip()

# The original plot for this example
dietswap %>%
  ps_filter(timepoint == 1) %>% 
  comp_barplot(
    tax_level = "Genus", n_taxa = 15, other_name = "Other",
    taxon_renamer = function(x) stringr::str_remove(x, " [ae]t rel."),
    palette = distinct_palette(n = 15, add = "grey90"),
    merge_other = FALSE, bar_outline_colour = "darkgrey"
  ) +
  coord_flip()