Creating a mirrorplot in R

Throughout the process of running an RNA-seq, you may be asked to graphically represent total number of genes that are upregulated and downregulated. A mirror plot allows you to do just that and can be included in presentations:

This is created with ggplot2 and labels are added using ggrepel.

Lets get into the specifics below.

Running from the files in repository

If you want to run from the files in the repository, download the mirrorplot.r file and run. It will create a mirror plot according to the parameters shown below.

Installing libraries

As mentioned before, we'll be using ggplot2 and ggrepel.

library(ggplot2)
library(ggrepel)

Getting the data ready

The first thing you'll want to setup are the major categories of tissue/cell types. In my example, I'm using plant tissue so:

sample.type1 <- c("Root", "Stem", "Leaf")

Next, lets assign values. This will just be a long list of values, two for each of the categories above. The first will be positive, and the second negative reflecting up/down regulation. Because we have three categories, there will be six total elements of this list.

sample.values1 <- c(500, -300, 200, -50, 200, -30) 

Then we want to setup two lists that contain the fill colors to be used in the plot as well as the text that is going to be displayed vertically.

bar.fill1 <- c("white", "black")
regulation.type1 <- c("Upregulated", "Downregulated")
label.text.color1 <- rev(bar.fill1)
maxnum <- max(sample.values1)
y.lab <- "Number of Genes" 

The text color will simply be the reversed values of the fill color since we want these to show up against the fill colors. The variable that I've setup as the maxnum for the dataset has an impact on the placement for the up/down regulation labels, and the y.lab is the label that will be used for the y axis.

Now, we can finally setup our dataframe

dat1 <- data.frame(sample.type=rep(sample.type1, each = 2), 
                  regulation.type = regulation.type1,
                  y = sample.values1, 
                  bar.fill = bar.fill1, 
                  label.text.color = label.text.color1, 
                  stringsAsFactors = FALSE)

dat1$regulation.type <- factor(dat1$regulation.type, levels = regulation.type1)
dat1$sample.type <- factor(dat1$sample.type, levels = sample.type1)

Will display as:

sample.type	regulation.type	y	bar.fill	label.text.color
Root	Upregulated	500	white	black
Root	Downregulated	-300	black	white
Stem	Upregulated	200	white	black
Stem	Downregulated	-50	black	white
Leaf	Upregulated	200	white	black
Leaf	Downregulated	-30	black	white

Setting the regulation.type and sample.type as ordered factors will allow them to appear in the order that you specified when you setup the lists.

Creating the plot

Now we finally get to setup our plot.

p <- ggplot(dat1, aes(x=sample.type, y=y, fill=regulation.type, label=abs(y))) + 
  geom_bar(stat="identity", position="identity", color = "black", width=.6)

The x axis is using our sample.type, the y axis the number of genes that are regulated, and we've set the future labels that will be used as the absolute values (so we won't be bothered with negative signs appearing in our plot).

Additionally, the type of chart is specified as a geom_bar and a black border.

Running the code produces this plot:

Now, there are several things that we need to do to this plot to make it appear as we'd like:

1. Adjust the fill colors
2. Relabel/remove some axes
3. Add the labels for each of the bar values
4. Remove the legend

p <- p + scale_fill_manual(values = dat1$bar.fill) + 
  theme(axis.title.x=element_blank(),axis.text.y=element_blank(), 
        axis.title.y = element_text(size=50, face="bold"), 
        axis.text=element_text(size=50, face="bold")) +
  ylab(y.lab) +
  geom_label_repel(direction = "y", mapping = aes(hjust=.5, 
                                                  vjust = rep(c(1,0), length(sample.type1))), 
                   color = dat1$label.text.color, 
                   fill= dat1$bar.fill, size=15, 
                   label.padding = unit(.5, "lines")) +
  theme(legend.position = "none")

Please note: the size of the font has been optimized for a 1200x1200 plot. If you are running this in RStudio, be sure to adjust the export size, otherwise, text will appear abnormally large in the preview window.

We're getting close, but you can see that we still have to add the vertical up/downregulation label, and the chart cuts off the labels for some of the larger values. We can fix that with:

p <- p + annotate(geom="text", x=.45, y=.02*maxnum, label=regulation.type1[1],
           color="black", angle=90, fontface="italic", size=12, hjust=0) +
  annotate(geom="text", x=.45, y=-.02*maxnum, label=regulation.type1[2],
           color="black", angle=90, fontface="italic", size=12, hjust =1) + 
  scale_y_continuous(expand=expansion(mult=c(.10,.10)))

As a finishing touch, lets add a horizontal line and call it a day!

p <- p + geom_hline(yintercept=0, color="black", size=2)

Conclusion

Sometimes, labels are desired to be in the actual bar. However, after creating many of these for an RNA-seq I found that because of scaling issues between samples, the label doesn't always fit within the bar. Because of that, I'm using labels outside of the bar using ggrepel.

Additionally, the script does allow the flexibility to add many different tissue types, however, my personal script includes the up/downregulated counts and names of tissues in their own list which makes it a bit easier to understand the structure:

sample.type1 <- list()
sample.type1[["Root"]] <- c(500,-300)
sample.type1[["Stem"]] <- c(200,-50) 
sample.type1[["Leaf"]] <- c(200,-30) 

maxnum <- 0
for (i in 1:length(sample.type1)){
    maxnum <- max(maxnum, 
    sample.type1[[i]])
}

Then integrating them into a new dataframe using a loop:

for(i in 1:length(sample.type1)){
	    intdf <- 	    data.frame(sample.type=rep(names(sample.type1[i]), 2), regulation.type=c("Upregulated", "Downregulated"), y=sample.type1[[i]], bar.fill = c("White", "Black"), label.text.color = c("Black","White"))
	    dat <- rbind(dat, intdf)
  }