ggplot: position dodge based on category

Question

I'd like to use position dodge to offset one variable in my ggplot chart (in this case banana) and leave the other two variables (red_apple and green_apple) without an offset. Using position_dodge applies the offset to each variable, but I'd like to choose which variables are offset specifically.

library(ggplot2)

data <- data.frame(Place = c(rep('Place_A',30),rep('Place_B',30)),
                   variable =     c(rep(c(rep('red_Apple',10),rep('green_Apple',10),rep('bananna',10)),2)),
                   value = rep(c(1:10,1:10-.05,1:10+.2),2))

dodge = position_dodge(.5)

ggplot(data, aes(Place, value)) + 
  geom_point(aes(color=variable),position=dodge)

Is there for example a way to scale position manually, like how you can do for other aesthetics? This obviously throws an error, but is what I was hoping for...

ggplot(data, aes(Place, value)) + 
  geom_point(aes(color=variable, position = variable)) + 
  scale_position_manual(breaks = c('green_Apple','red_Apple','bananna'),
                      values = c(position_dodge(0),position_dodge(0),position_dodge(.5)))

Answer 1

Does this look like what you want?

data$grp = ifelse(data$variable == "bananna", 2, 1)
ggplot(data, aes(Place, value, group = grp)) + 
  geom_point(aes(color=variable), position = position_dodge(0.5))