Setting the axes of a ggplot scatterplot to manually line up with geom_rect elements

I'm trying to make a scatter plot with a log scale that has 3 geom_rect() features laying out different "bins" for some density features.

Whenever I add the geom_rect() features, it increases the limits of the x and y axis. Ideally, I'd like the rectangles to line up with the maximum x and y value displayed on the chart, removing all white space. Whenever I tried to set the limits on the axes using the following plot elements, it just results in an error:

> Error in seq.default(min, max, by = by) : 'from' must be a finite number"

  scale_x_continuous(trans='log2', limits = c(0, 8000)) +
  scale_y_continuous(trans='log2', limits = c(0, 8000)) +

Here's an example of the chart:

And here's the code that was used in its production.

ggplot(final_df, aes(x=as.numeric(stroma), y=as.numeric(epi), 
                     label = ID)) +
  geom_rect(aes(xmin=0, xmax=stroma_cutoff, ymin=0, ymax = epi_cutoff),
            alpha = 0.005, fill = "blue2") +
  geom_rect(aes(xmin=stroma_cutoff, xmax=8000, ymin=0, ymax = epi_cutoff),
            alpha = 0.005, fill = "yellow1") +
  geom_rect(aes(xmin=0, xmax=8000, ymin=epi_cutoff, ymax = 4000),
            alpha = 0.005, fill = "red1") +
  geom_point(size = 2) +
  geom_point(data = crc_vals,
             aes(x=as.numeric(stroma), y=as.numeric(epi)),
             color = 'green',
             size = 3) +
  geom_point(data = crc_met_vals,
             aes(x=as.numeric(stroma), y=as.numeric(epi)),
             color = 'purple',
             size = 3) +
  scale_x_continuous(trans='log2') +
  scale_y_continuous(trans='log2') +
  labs(y='Feature1',x='Feature2',
       title = 'Chart\n
       blue2 0.005 - yellow1 0.005 - red1 0.005') +
  theme_bw()

And here's the structure of the dataframes used.

> dput(head(final_df))
structure(list(ID = c("USL-2022-41628", "USL-2022-41637", "USL-2022-41640", 
"USL-2022-41659", "USL-2022-41666", "USL-2022-41669"), stroma = c(639.7364171, 
78.23842193, 484.908185, 263.4609547, 159.693075, 18.38209375
), epi = c(424.6941624, 55.78718892, 103.9890452, 270.4756807, 
47.79603601, 12.29226541), Indication = c("Gastric", "Gastric met", 
"Gastric met", "Esophageal met", "Gastric", "Gastric met")), na.action = structure(c(`USL-2022-42047` = 106L), class = "omit"), row.names = c(NA, 
6L), class = "data.frame")
> dput(head(crc_vals))
structure(list(ID = c("USL-2022-41663", "USL-2022-41696", "USL-2022-41698", 
"USL-2022-41727", "USL-2022-41735"), stroma = c(842.3719975, 
283.5923779, 1073.589824, 306.987757, 1217.564582), epi = c(144.1873575, 
37.71616383, 338.2043632, 52.77961909, 410.411979), Indication = c("CRC", 
"CRC", "CRC", "CRC", "CRC")), na.action = structure(c(`USL-2022-42047` = 106L), class = "omit"), row.names = c(48L, 
53L, 54L, 59L, 61L), class = "data.frame")
> dput(head(crc_met_vals))
structure(list(ID = c("USL-2022-41665", "USL-2022-41667", "USL-2022-41668", 
"USL-2022-41701", "USL-2022-41729"), stroma = c(331.6539946, 
2398.683215, 184.7736286, 8.277558758, 435.3299529), epi = c(153.3781584, 
2022.155922, 10.2111996, 9.725953771, 61.15902551), Indication = c("CRC met", 
"CRC met", "CRC met", "CRC met", "CRC met")), na.action = structure(c(`USL-2022-42047` = 106L), class = "omit"), row.names = c(49L, 
50L, 51L, 55L, 60L), class = "data.frame")

Both epi_cutoff and stroma_cutoff are 200.

To remove the whitespace, you need to use expand(c(0,0) in your scale_... functions.

Since you have log-scale axes, lower limit cannot be set to 0. You can use limits = c(1, 8000) if you want to, but that has (almost) nothing to do with the whitespace. Moreover, since your "top" geom_rect only goes to ymax = 4000 I would not set the limit for y to 8000. Otherwise, you end up with whitespace again.

I actually used limits = c(0.1, 8000) for x-axis and limits = c(0.1, 2^12) for y-axis (while changing ymax to 2^12) to get cleaner axes ticks and avoid any whitespace.

Keep in mind that if you have negative values or zeros, then you should transform your data prior to plotting. Otherwise, you'll loose those data points (see the warnings shown below after plotting). You can add 1 to abs() of each value. Then use different shapes to differentiate between negative values, positive ones, and zeros. Axes tick labels can be manipulated to show the right values without that added 1.

library(ggplot2)

ggplot(final_df, aes(x=as.numeric(stroma), y=as.numeric(epi), 
                     label = ID)) +
  geom_rect(aes(xmin=0, xmax=stroma_cutoff, ymin=0, ymax = epi_cutoff),
            alpha = 0.1, fill = "blue2") +
  geom_rect(aes(xmin=stroma_cutoff, xmax=8000, ymin=0, ymax = epi_cutoff),
            alpha = 0.1, fill = "yellow1") +
  geom_rect(aes(xmin=0, xmax=8000, ymin=epi_cutoff, ymax = 2^12),
            alpha = 0.1, fill = "red1") +
  geom_point(size = 2) +
  geom_point(data = crc_vals,
             aes(x=as.numeric(stroma), y=as.numeric(epi)),
             color = 'green',
             size = 3) +
  geom_point(data = crc_met_vals,
             aes(x=as.numeric(stroma), y=as.numeric(epi)),
             color = 'purple',
             size = 3) +
  scale_x_continuous(trans='log2',  expand = c(0,0), limits = c(0.1, 8000)) + 
  scale_y_continuous(trans='log2',  expand = c(0,0), limits = c(0.1, 2^12)) + 
  labs(y='Feature1',x='Feature2',
       title = 'Chart\n
       blue2: 0.1 - yellow1: 0.1 - red1: 0.1') +
  theme_bw()
#> Warning: Transformation introduced infinite values in continuous x-axis
#> Warning: Transformation introduced infinite values in continuous y-axis
#> Transformation introduced infinite values in continuous y-axis
#> Warning: Transformation introduced infinite values in continuous x-axis

<!-- -->

^{Created on 2023-05-22 with reprex v2.0.2}