R dplyr heatmap dealing with missing values

Question

so I have run into a minor issue with drawing heat maps using geom_tile() and dplyr package. I assume that it is a simple solution, but I have not been able to find the answer anywhere yet. Apologies if there is one out there and I just missed it.

So the following code is for a trivial example:

Trivial <- tibble(
  Name1 = c("a","b","c"),
  Name2 = c("x","y","z"),
  Value = c(1,2,3)
)
Trivial %>%
  ggplot(aes(Name1, Name2)) +
    geom_tile(aes(fill = Value)) +
    scale_fill_gradient(low = "white", high = "green")

The heat map produced by this shows grey/blank space where there is no information for those combinations of Name1 and Name2. Is there a way of instead of showing blank space I could fill it with white (or any other colour)? Alternatively, could I put a "NA" on the plot to show that those blank spaces are indeed meant to be blank?

I have tried using the na.value argument in scale_fill_gradient, but it does not work. I assume since there are no NAs in the table. However, I think that could be a way forward. I will keep experimenting, but any help would be greatly appreciated!

Thank you!

Answer 1

Or you can just add them_classic() .

Trivial %>%
    ggplot(aes(Name1, Name2)) +
    geom_tile(aes(fill = Value)) +
    scale_fill_gradient(low = "gray", high = "green") + 
    theme_classic()

I just used low= "gray" because background is white.

You can change also the background :

Trivial %>%
    ggplot(aes(Name1, Name2)) +
    geom_tile(aes(fill = Value)) +
    scale_fill_gradient() + 
    theme_classic() +
    theme(panel.background = element_rect(fill = "white", colour = "white"))

Answer 2

You can use expand() from tidyr to get all combinations of Name1 and Name2 and then fill all NA with 0 and then plot the heatmap:

library(tidyr)      
Trivial %>% expand(Name1, Name2) %>% 
  left_join(Trivial, by = c("Name1", "Name2")) %>%
  mutate( Value = ifelse(is.na(Value), 0, Value)) %>%
  ggplot(aes(Name1, Name2)) +
  geom_tile(aes(fill = Value)) +
  scale_fill_gradient(low = "white", high = "green")

The result looks like this: