Set / Link point and shape options for variables in ggplot2-CodePudding

I would like to link variables I have in a dataframe i.e. ('prop1', 'prop2', 'prop3') to specific colours and shapes in the plot. However, I also want to exclude data (using dplyr::filter) to customise the plot display WITHOUT changing the points and shapes used for a specific variable. A minimal example is given below.

  library(ggplot2)
  library(dplyr)
  library(magrittr)

obj <- c("cmpd 1","cmpd 1","cmpd 1","cmpd 2","cmpd 2")
  x <- c(1, 2, 4, 7, 3)
  var <- c("prop1","prop2","prop3","prop2","prop3")
  y <- c(1, 2, 3, 2.5, 4)
  col <- c("#E69F00","#9E0142","#56B4E9","#9E0142","#56B4E9")
  shp <- c(0,1,2,1,2)
  df2 <- cbind.data.frame(obj,x,var,y,col,shp)

  plot <- ggplot(data = df2 %>%
                       filter(obj %in% c(
                         "cmpd 1",
                         "cmpd 2"
                       )),
                     aes(x = x,
                         y = y,
                         colour = as.factor(var), 
                         shape = as.factor(var)))    
    geom_point(size=2)   
    #scale_shape_manual(values=shp)  
    #scale_color_manual(values=col)   
    facet_grid(.~obj) 
  plot

However, when I redact cmpd1 (just hashing in code) the colour and shape of prop2 and prop3 for cmpd2 change (please see plot2).

To this end, I tried adding in scale_shape_manual and scale_color_manual to the code (currently hashed) and linked these to specific vars (col and shp) in the dataframe (df2), but the same problem arises that both the shape and color of these variables changes when excluding one of the conditions?

Any and all help appreciated.

CodePudding user response：

Try something like this:

library(tidyverse)
obj <- c("cmpd 1","cmpd 1","cmpd 1","cmpd 2","cmpd 2")
x <- c(1, 2, 4, 7, 3)
var <- c("prop1","prop2","prop3","prop2","prop3")
y <- c(1, 2, 3, 2.5, 4)

df2 <- cbind.data.frame(obj,x,var,y)
col <- c("prop1" = "#E69F00", 
         "prop2" = "#9E0142",
         "prop3" = "#56B4E9")
shp <- c("prop1" = 0, 
         "prop2" = 1,
         "prop3" = 2)

plot <- ggplot(data = df2 %>%
                 filter(obj %in% c(
                   "cmpd 1",
                   "cmpd 2"
                 )),
               aes(x = x,
                   y = y,
                   colour = var, 
                   shape = var))    
  geom_point(size=2)   
  scale_shape_manual(values=shp)  
  scale_color_manual(values=col)   
  facet_grid(.~obj) 
plot