How to set the color value in a plot based on 2 conditions ggplot

I am trying to make a grouped barchart. I’ve reused an older script that worked before, however in that script i didn’t have to color based on conditions. This is what I am trying now and failing at.
I want the bars to be coloured based on the Condition and if the value is below or above 0
Example data:

library( ggplot2)
library(reshape2)
library(Rcpp)


 prot_names <- c("Prot_A", "Prot_B","Prot_C")
Expression_Con_A <- c(-1,2,1)
Expression_Con_B <- c(-2,1.5,0.5)

df <- data.frame(prot_names, Expression_Con_A, Expression_Con_B)
melt_dat <- melt((df))

I’m imagining something like:

for (i in melt_dat$value) {
  if (melt_dat$value[i] > 0 & melt_dat$variable== "ConA")
  {melt_dat$color<- skyblue1}
  else if (melt_dat$value[i] < 0 & melt_dat$variable== "ConA")
  {melt_dat$color<- red1}
  else if (melt_dat$value[i] > 0 & melt_dat$variable== "ConB")
  {melt_dat$color<- skyblue4}
  else if (melt_dat$value[i] < 0 & melt_dat$variable== "ConB")
  {melt_dat$color<- red4}
}

However this gives me the folowing error:
Error in if (melt_dat$value[i] > 0 & melt_dat$variable == "ConA") { :
the condition has length > 1

ggplot2: how to color a graph by multiple variables
This worked in a way but it gave me my bars in a colour and the frame in another which is not what I want

I also tried putting in the colors by hand

g<-ggplot(melt_dat, aes(x=value, y=prot_names, fill=variable))+
            geom_bar(stat='identity', position='dodge')+
            scale_fill_manual(values = c("red1","skyblue1","red1","red4","skyblue4","red4")

which did not work either as it only colours by condition (only uses red1 and skyblue 1 independant of value>0)

>Solution :

You can use nested ifelse (or if_else), but in case of multiple conditions it’d be easier to use case_when from dplyr, In your case:

library(dplyr)

melt_dat <- melt_dat %>% mutate("type" = case_when(value > 0 & variable == "Expression_Con_A" ~ "color1",
                                       value < 0 & variable == "Expression_Con_A" ~ "color2",
                                       value > 0 & variable == "Expression_Con_B" ~ "color3",
                                       value < 0 & variable == "Expression_Con_B" ~ "color4"))

ggplot(melt_dat, aes(x=value, y=prot_names, fill=type)) + 
  geom_bar(stat='identity', position='dodge') + 
  scale_fill_manual(values = c("red1","skyblue1","red1","red4","skyblue4","red4"))

I defined a new variable "type", and note that in the ggplot you’ve to change the "fill" argument from "variable" to this new variable.