How to remove unused values from ggplot2 scale_color_manual()?

Question

I think it is a recent change in ggplot2 that all the values defined in the scale_manual function are included in the legend, but I want only the values that are in the data set in the legend.

Below is a minimal example, in which "C" is filtered from the data set but still appears in the legend. Is there an easy fix to remove "C" from the legend?

library(ggplot2)
library(dplyr)

df_dummy <- tribble(
    ~label, ~x, ~y,
    "A",     1,  1,
    "B",     2,  2,
    "C",     3,  3,
    "D",     4,  4
)

myColors <- viridis::viridis_pal(option = "H")(4)
names(myColors) <- factor(df_dummy$label)
dummy_col_scale <- scale_color_manual(name = "Legend", values = myColors)

df_dummy %>% 
    filter(label != "C") %>% 
    ggplot(aes(x, y, color = label)) +
    geom_point() +
    dummy_col_scale

sessionInfo()

R version 4.0.3 (2020-10-10)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19044)

Matrix products: default

locale:
[1] LC_COLLATE=English_Australia.1252 
[2] LC_CTYPE=English_Australia.1252   
[3] LC_MONETARY=English_Australia.1252
[4] LC_NUMERIC=C                      
[5] LC_TIME=English_Australia.1252    

attached base packages:
[1] stats     graphics  grDevices utils    
[5] datasets  methods   base     

other attached packages:
[1] dplyr_1.0.5   ggplot2_3.3.5

loaded via a namespace (and not attached):
 [1] rstudioapi_0.13   magrittr_2.0.1   
 [3] tidyselect_1.1.1  munsell_0.5.0    
 [5] cowplot_1.1.1     viridisLite_0.4.0
 [7] colorspace_2.0-2  R6_2.5.1         
 [9] rlang_0.4.10      fansi_0.4.2      
[11] tools_4.0.3       grid_4.0.3       
[13] gtable_0.3.0      utf8_1.2.1       
[15] cli_3.0.1         DBI_1.1.1        
[17] withr_2.4.2       ellipsis_0.3.2   
[19] digest_0.6.27     assertthat_0.2.1 
[21] tibble_3.1.1      lifecycle_1.0.1  
[23] crayon_1.4.1      gridExtra_2.3    
[25] farver_2.1.0      purrr_0.3.4      
[27] viridis_0.6.0     vctrs_0.3.8      
[29] glue_1.4.2        labeling_0.4.2   
[31] compiler_4.0.3    pillar_1.6.4     
[33] generics_0.1.0    scales_1.1.1     
[35] pkgconfig_2.0.3  

Answer 1

One option is to set limits = force in scale_color_manual.

df_dummy <- structure(list(label = c("A", "B", "C", "D"), x = c(1, 2, 3, 
4), y = c(1, 2, 3, 4)), class = c("tbl_df", "tbl", "data.frame"
), row.names = c(NA, -4L))

myColors <- viridis::viridis_pal(option = "H")(4)
names(myColors) <- factor(df_dummy$label)

dummy_col_scale <-
  scale_color_manual(name = "Legend",
                     values = myColors,
                     limits = force)

df_dummy %>%
  filter(label != "C") %>%
  ggplot(aes(x, y, color = label)) +
  geom_point() +
  dummy_col_scale

Output

Answer 2

I believe there are a few solutions to your problem. One potential solution is to do the filtering first then select 3 colours:

library(ggplot2)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

df_dummy <- tribble(
  ~label, ~x, ~y,
  "A",     1,  1,
  "B",     2,  2,
  "C",     3,  3,
  "D",     4,  4
)

df2 <- df_dummy %>% 
  filter(label != "C")

myColors <- viridis::viridis_pal(option = "H")(3)
names(myColors) <- factor(df2$label)
dummy_col_scale <- scale_color_manual(name = "Legend", values = myColors)

df2 %>%
  ggplot(aes(x, y, color = label)) +
  geom_point() +
  dummy_col_scale

Another option is to set limits in dummy_col_scale, i.e.

df_dummy <- tribble(
  ~label, ~x, ~y,
  "A",     1,  1,
  "B",     2,  2,
  "C",     3,  3,
  "D",     4,  4
)

df2 <- df_dummy %>% 
  filter(label != "C")

myColors <- viridis::viridis_pal(option = "H")(4)
names(myColors) <- factor(df2$label)
dummy_col_scale <- scale_color_manual(name = "Legend", values = myColors,
                                      limits = c("A", "B", "D"))

df2 %>%
  ggplot(aes(x, y, color = label)) +
  geom_point() +
  dummy_col_scale

^{Created on 2021-12-03 by the reprex package (v2.0.1)}