ggplot frienndly corrlation plots with ggcorrplot

A co-worker ran into issues with corrplot::corrplot() cutting off the title. A useful alternative is ggcorrplot. It makes okay plots with ggplot2’s logic. Not as clean as the above but it’ll work with patchwork and cowplot. Unfortunately, scale_colour_stepsn doesn’t override the scaling.

library(ggcorrplot)

p.mat <- cor_pmat(cor_df)
ggcorrplot(cor(cor_df, use = "pairwise.complete.obs"), 
           p.mat = p.mat,
           insig = "blank",
           type = "upper",
           outline.col = "white",
           colors = RColorBrewer::brewer.pal(n = 9, name = "PuOr")[c(1,5,9)]
           )+
  labs(title = "Brian_AP_Delayed")

ggcorrplot appears to call internal functions which makes modifying it quickly impractical (one would probably be best forking the package and modifying that). I think I have a workaround that gets the same binning behavior:

After the significance matrix (p.mat) is generated overwrite the correlation matrix with the middle value of each desired bin.

          bkkca      cav1      cav2
bkkca 1.0000000 0.3452702 0.5603564
cav1  0.3452702 1.0000000 0.7880727
cav2  0.5603564 0.7880727 1.0000000
> # bin the correlations so there are fewer colors used in the figure
> cor_bins <- seq(-1, 1, length.out = 9)
> for (i in 1:(length(cor_bins)-1)){
+   test[test > cor_bins[i] &amp; test < cor_bins[i+1]] <- ((cor_bins[i] + cor_bins[i+1])/2)
+ }
> test
      bkkca  cav1  cav2
bkkca 1.000 0.375 0.625
cav1  0.375 1.000 0.875
cav2  0.625 0.875 1.000

Here this makes very slight changes to the plot. (legend dropped to not imply a continuous fill)

(2021-2-2) Last update, this is harder to read up will use the more extreme value to get closer to corrplot

test <- seq(-1, 1, length.out = 5)+.0000001
test
#-0.9999999 -0.4999999  0.0000001  0.5000001  1.0000001
for (i in 1:(length(cor_bins)-1)){
  # test[test > cor_bins[i] &amp; test < cor_bins[i+1]] <- (cor_bins[i] + cor_bins[i+1])
  test[test > cor_bins[i] &amp; test < cor_bins[i+1]] <-   cor_bins[c(i, (i+1))[which.max(abs(cor_bins[i:(i+1)]))]]
}
test
#-1.00 -0.50  0.25  0.75  1.00