R：使用带有geom_tile（）和scale_fill_gradientn（）的颜色划分热图

Question

I am producing many heat maps which should all have the same color scheme, as follows:

• if x-axis value = y-axis value, fill with blue gradient
• if x-axis value > y-axis value, fill with yellow gradient
• if x-axis value < y-axis value , fill with red gradient

EDIT: It is important that each color gradients over their respective numbers. Ie in Example A below, 18, 20, and 25 are the "darkest" colors on the map.

My code is below:

library(ggplot2)
library(dplyr)
library(scales)
graphdata$pct <-ifelse(graphdata$A-graphdata$B>=0,graphdata$n/sum(graphdata$n),-graphdata$n/sum(graphdata$n))
graphdata$color_group <- ifelse(graphdata$A-graphdata$B<0,1,
                                   ifelse(graphdata$A-graphdata$B>0,2,3))
    graphdata$rescale <- 100*graphdata$pct + 100*(graphdata$color_group-1)
    gradientends <-c(min(graphdata$rescale),
                     ifelse(dim(subset(graphdata,rescale<=0,select=c(rescale)))[1]==0,0,max(subset(graphdata,rescale<=0,select=c(rescale)))),
                     ifelse(dim(subset(graphdata,rescale>=100 & rescale<200,select=c(rescale)))[1]==0,0,min(subset(graphdata,rescale>=100 & rescale<200,select=c(rescale)))),
                     ifelse(dim(subset(graphdata,rescale>=100 & rescale<200,select=c(rescale)))[1]==0,0,max(subset(graphdata,rescale>=100 & rescale<200,select=c(rescale)))),
                     ifelse(dim(subset(graphdata,rescale>=200,select=c(rescale)))[1]==0,0,min(subset(graphdata,rescale>=200,select=c(rescale)))),
                     max(graphdata$rescale))

    colorends <- c("tomato1","lightpink1","lightgoldenrod1","goldenrod1","lightsteelblue1","steelblue2")

    ggplot(data = graphdata, aes(x = as.factor(A), y = as.factor(B))) +
          geom_tile(aes(fill = rescale), colour = "white") + 
          scale_fill_gradientn(colors=colorends,values=rescale(gradientends)) +
          theme(legend.position = "none", 
                axis.text=element_text(size=12,face="bold"), 
                panel.background = element_blank(),
                axis.ticks=element_blank(), 
                plot.title = element_text(hjust=0.5)) + 
          geom_text(aes(label=n)) +labs(x="A",y="B")

My code works when there is data that can be bucketed in all three categories (Example A), but it fails to color correctly when there are 2 or less categories in the data (Example B).

Example A

A <- c(1,2,3,1,2,3,2,1,3)
B <- c(1,2,3,2,3,2,1,3,1)
n <- c(10,5,20,15,18,10,10,10,25)

graphdata <- data.frame(A,B,n)
rm(A,B,n)

gives me what I'm expecting:

Example B

A <- c(1,2,3)
B <- c(1,2,3)
n <- c(10,5,20)

graphdata <- data.frame(A,B,n)
rm(A,B,n)

does not give me a blue gradient:

I would like Example B to look more like this:

Answer 1

I would advise using alpha to scale your colors. This makes the problem much simpler:

cols <- c('-1' = "tomato1", '0' = "steelblue2", '1' = "goldenrod1")

ggplot(graphdata, aes(x = as.factor(A), y = as.factor(B))) +
    geom_tile(aes(fill = as.factor(sign(A-B)), alpha = n), colour = "white") + 
    scale_fill_manual(values = cols) +
    geom_text(aes(label = n)) +
    labs(x = "A", y = "B") +
    theme_minimal() + 
    theme(legend.position = 'none', panel.grid = element_blank())