ggplateplot

Well, well, well… it seems you’ve found this package for plotting: wells.

The goal of {ggplateplot} is to extend {ggplot2} to make it easier to make plots resembling cell culture microwell plates. This packages takes care of some of the layouts, relying on the flexibility of {ggplot2} and extensions to add layers, scales, theme adjustments et cetera.

Installation

You can install the development version of {ggplateplot} like so:

# install.packages("remotes")
remotes::install_github("teunbrand/ggplateplot")

Example

This is a terse example of how one could make a plot from a matrix of values.

library(ggplot2)
library(ggplateplot)

plate <- matrix(rnorm(24), nrow = 4, ncol = 6)

ggplateplot(plate, aes(fill = value)) +
  geom_well()

Instead of providing a matrix, we can also provide long-format data. It should automatically detect an appropriate layout for the plate based on the number of rows in the data argument, when it has 6, 12, 24, 48, 96 or 384 rows.

df <- expand.grid(
  x = 1:8,
  y = LETTERS[1:6]
)
df$values <- rnorm(nrow(df))

ggplateplot(df, aes(x, y, fill = values)) +
  geom_well()

How does it work?

This package is not a package that instantly makes your plots prettier or easier. It is a formatting tool for a very particular kind of plot, of microwell shaped data, in {ggplot2}’s ecosystem.

The ggplateplot() function above is a thin wrapper around the ggplot() function. It does a few automated things to make your plots look more like cell culture plates.

Geom

First off, the {ggplateplot} package has a specialised layer (geom) for drawing wells. This isn’t a particularly exciting layer as it just draws circles as you would with geom_point(). A small bit of magic comes from combining this geom with coord_plate(), in that the size aesthetic is synced with the diameter of wells.

p <- ggplot(df, aes(x, y, fill = values)) +
  geom_well()
p

Coord

The crux of {ggplateplot} is coord_plate(), which modifies how several panel components of a plot are drawn.

Theming

The following adjustments to the theming of a plot are made.

The panel.background and panel.border follow the shape of a microwell plate. Corners are rounded a little bit and you can ‘bite’ off some corners with the corner argument to resemble a microwell plate better.
The panel.grid.major now draws circles around the wells.
The panel.grid.minor performs the role of the major panel grid.

p + coord_plate(corner = "topright", spec = 48) +
  theme(
    panel.grid.major = element_line(colour = "dodgerblue"),
    panel.grid.minor = element_line(colour = "tomato"),
    panel.background = element_rect(colour = "black", fill = "white")
  )

Layout

The coord_plate() function has a specs argument that takes a description of a well plate. We’ve included descriptions for 6, 12, 24, 48, 96 and 384-well plates in the package. If we mismatch the number of datapoints with the layout, we might get inappropriate layouts. In the example below, we’re purposfully giving the wrong layout for the shape of the data to demonstrate that, among other things, the spacing of wells is off.

p + coord_plate(specs = 96)

To offer slightly more flexibility than the standard layouts, you can set a custom layout with custom_plate_spec().

df <- data.frame(
  x = rep(1:5, each = 4),
  y = rep(LETTERS[1:4], 5),
  value = rnorm(20)
)

ggplot(df, aes(x, y, fill = value)) +
  geom_well() +
  coord_plate(spec = custom_plate_spec(ncol = 5, nrow = 4))

If you want to even further customise the layout, you can use new_plate_spec() to control even more parameters of the layout.

spec <- new_plate_spec(
  width = 100, height = 100,
  hor_spacing = 15,
  ver_spacing = 20,
  well_diameter = 12, 
  ncol = 5, nrow = 4,
  corner_size = 25
)

ggplot(df, aes(x, y, fill = value)) +
  geom_well() +
  coord_plate(spec = spec)

Circling back

So what does ggplateplot() do exactly? Well, it just tries to automatically match the shape of your data to an appropriate setting for coord_plate(). If you have data that doesn’t match the dimensions of one of the standard plates, we recommend that you simply use ggplot() + coord_plate() instead, and customise from there. For matrix input, it additionally converts it to long format and automatically sets x and y aesthetics. The values in the matrix are converted to a value column.

plate <- matrix(rnorm(384), ncol = 24, nrow = 16)

ggplateplot(plate, aes(fill = value)) +
  geom_well()

Interacting with ggplot2

While {ggplateplot} makes it easier to make plots of plates, it still requires some familiarity with the ggplot2 package. The ggplateplot() functions return regular ggplot objects, so they can be combined with the usual functions one can use to manipulate plots. Below is an example of combining it with other {ggplot2} functions. We can use an array instead of matrix input to make variable out of the 3rd dimension, which we can use to create facets.

m <- array(1:192, dim = c(8, 12, 2))

ggplateplot(m, aes(fill = value)) +
  geom_point(shape = 24, size = 2.5) +
  scale_fill_viridis_c() +
  scale_y_reverse(name = "Rows", labels = LETTERS[1:8]) +
  facet_wrap(~ Var3) +
  theme(
    panel.background = element_rect(colour = "black", fill = NA),
    panel.grid.major = element_line(colour = "grey90"),
    panel.grid.minor = element_line(colour = "grey97")
  ) +
  labs(x = "Columns")

teunbrand / ggplateplot