Use matrix view for axis labels for an existing ggplot2 plot?

[dhimmel]

I'm creating a barplot visualization to show compare three methods of accessing scholarly articles. In this plot, there are three methods: oaDOI, Penn, and Sci-Hub. The plot shows the coverage of the methods independently as well as possible combinations:

I want to use UpSetR's matrix view for the axis labels. Is there anyway to apply the matrix view to an existing ggplot2 plot axis? I'm interested in the following three things:

1. Using ggplot2's syntax to define the plot as much as possible (i.e. not having to reimplement themes, ggplot2::geom_col, ggplot2::geom_text, etcetera).
2. Changing the axis to use a matrix view
3. Applying this to faceted plots (i.e. with ggplot2::facet_grid, 1 & 2 are bigger priorities).

Is this possible?

Expand to see code and data for example

Here's the R code to create the plot above:

`%>%` = dplyr::`%>%`
plot_df %>%
  ggplot2::ggplot(ggplot2::aes(x = repos, y = coverage)) +
  ggplot2::geom_col(fill='pink') +
  ggplot2::geom_text(ggplot2::aes(label = label, y = 0.02), size=2.5, hjust='inward', color='#000000') +
  ggplot2::scale_x_discrete(name = NULL, expand = c(0.02, 0)) +
  ggplot2::scale_y_continuous(name = "Repository's Coverage", labels = scales::percent, expand = c(0, 0), breaks=seq(0, 1, 0.2)) +
  ggplot2::expand_limits(y = 1) +
  ggplot2::coord_flip() +
  ggplot2::theme_bw() +
  ggplot2::theme(
    panel.grid.major.y = ggplot2::element_blank(),
    plot.margin = ggplot2::margin(t = 0, r = 12, b = 0, l = 5, unit='pt'))

Here is the contents of plot_df:

collection	oadoi_color	venn	repos	n_repos	available	articles	coverage	label
Combined	closed + green	100	oaDOI	1	25981	208786	0.12444	26K of 209K articles (12.4%)
Combined	closed + green	010	Penn	1	174375	208786	0.83519	174K of 209K articles (83.5%)
Combined	closed + green	001	Sci-Hub	1	189269	208786	0.90652	189K of 209K articles (90.7%)
Combined	closed + green	110	oaDOI, Penn	2	176915	208786	0.84735	177K of 209K articles (84.7%)
Combined	closed + green	101	oaDOI, Sci-Hub	2	191321	208786	0.91635	191K of 209K articles (91.6%)
Combined	closed + green	011	Penn, Sci-Hub	2	200496	208786	0.96029	200K of 209K articles (96.0%)
Combined	closed + green	111	oaDOI, Penn, Sci-Hub	3	201259	208786	0.96395	201K of 209K articles (96.4%)

[ngehlenborg]

Thanks @dhimmel! Could you maybe post a rough sketch of what you are trying to do? I am not sure if I am 100% sure that I understand what you are trying to achieve.

[dhimmel]

Hey @ngehlenborg, thanks for the help! The following modification of the above figure should demonstrate what I'm envisioning:

I want to replace the axis tick labels with an UpSetR matrix/dot view. Each bar in the plot represents the coverage of a certain access method or combination thereof.

Currently, I'm using a Venn-like diagram to show this (notebook):

However, these plots confuse people because they're not visualizing set overlap between the three access methods. Instead they show the coverage achieved by combining methods.

Essentially we want to use UpSetR to represent combinations of categories on a plot axis. Ideally, we'd continue using ggplot2 for the rest of the plot specification.