R: Heatmap with colour based on groups, NA values in grey and characters included

Question

I am trying to create a heatmap using ggplot and geom_tile. The fill colour is based on my x values and alpha based on values. Based on small example (left), I would like my Plot to look similar to this example (right). Two problems:

1. I get errors for the character elements on my alpha scale - is it possible to treat them like NA/ ignore them?
2. Actual NA values are coloured in the same colour as the group they belong to, rather than all getting a grey fill.

The real Data is much larger and contains facets.. hope this doesn't mess up any possible solution. Here is my example Dataset and my (semi-working) code:

X <- rep(st, each=3)
Y <- rep(st, times=3)

Values<- c('Apple', 2,3,NA, "Banana", 3,1,2,"Pear")

Data <- data.frame(X,Y,Values)

ggplot(Data, mapping = aes(x=X, y=Y, fill=X, 
                           alpha=Values # excluding this part I get a result, just not as I want it
       )) + 
  geom_tile(colour="white") +
  ylab("Y") + 
  xlab("X")+
  scale_fill_manual("Assay", 
                    values = c( 'red', 'yellow', 'green'),
                    na.value = 'grey')+
  scale_alpha("Value", na.value = 0.02)+
  ggtitle("Results Summary")+       
  theme( strip.text.y.left = element_text(angle = 0))+
  geom_text(label=Data$Values)

Thanks in advance for any help :)

Answer 1

Your first issue can be solved by converting Values to a numeric, i.e. mapping as.numeric(Values) on alpha.

Concerning the second issue. As you map X on fill the tiles are colored according to X. If you want to fill NAs differently as well as tiles where X==Y then you have to define your fill colors accordingly. To this end my approach adds a column fill to the df and makes use of scale_fill_identity.

Note that I moved the alpha and fill into geom_tile so that these are not passed on to geom_text...

... and following the suggestion by @AllanCameron I reversed the order of `Y' so that the plot is in line with your desired output.

library(ggplot2)
library(dplyr)

X <- rep(c('Apple', "Banana", "Pear"), each=3)
Y <- rep(c('Apple', "Banana", "Pear"), times=3)
Y <- factor(Y, levels = c("Pear", "Banana", "Apple"))

Values<- c('Apple', 2,3,NA, "Banana", 3,1,2,"Pear")

Data <- data.frame(X,Y,Values)
Data <- Data %>% 
  mutate(fill = case_when(
    is.na(Values) ~ "grey",
    X == Y ~ "white",
    X == "Apple" ~ "red",
    X == "Banana" ~ "yellow",
    X == "Pear" ~ "green"
  ))

ggplot(Data, mapping = aes(x=X, y=Y)) + 
  geom_tile(aes(fill=fill, alpha=as.numeric(Values)), colour="white") +
  ylab("Y") + 
  xlab("X")+
  scale_fill_identity("Assay") +
  scale_alpha("Value")+
  ggtitle("Results Summary")+       
  theme(strip.text.y.left = element_text(angle = 0))+
  geom_text(aes(label=if_else(!is.na(Values), Values, "NA")))
#> Warning in FUN(X[[i]], ...): NAs introduced by coercion