CODEWITHSUNDEEP
rggplot2reshape2
user3741035
user3741035

Reputation: 2545

combine facet_grid and boxplot

I want to make a facet-grid combined with boxplots. The facets should be the variable Experiment The boxplot should show value against FC for every variable.The variable and Experiment has multiple levels. Do I need to use gridExtra? I tried this, which is obviously wrong:

df.m

    FC          Experiment     variable value
1  -1.36811678 Ago2 Knockout     Pos1    no
2  -0.59630399 Ago2 Knockout     Pos1   yes
3  -0.09747337 Ago2 Knockout     Pos1    no
4   1.71652821 Ago2 Knockout     Pos1   yes
5  -1.13473546 Ago2 Knockout     Pos1    no
6  -0.44012950 Ago2 Knockout     Pos1   yes


ggplot(df.m,aes(value,FC,fill=as.factor(value))) + 
  geom_boxplot(notch=T) + facet_grid(~Experiment)

Upvotes: 1

Views: 11826

Answers (1)

hrbrmstr
hrbrmstr

Reputation: 78852

You can simulate some data if there's too much to post (and, knowing the levels of the various factors would also be helpful), but you were on the right track for the plot, you just needed to add an interaction for variable and value. I did the following with facet_wrap but you could just as easily use facet_grid:

library(ggplot2)

# create reproducible simulated data
set.seed(1492)
dat <- data.frame(FC=runif(100, min=-2, max=2),
                  Experiment=rep(c("Ago1", "Ago2", "Ago3", "Ago4"), 25),
                  variable=rep(c("Pos1", "Pos2", "Pos3", "Pos4", "Pos5"), 20),
                  value=sample(c("yes", "no"), 100, replace=TRUE),
                  stringsAsFactors=FALSE)

gg <- ggplot(dat, aes(x=interaction(variable, value), y=FC))
gg <- gg + geom_boxplot(aes(fill=value))
gg <- gg + facet_wrap(~Experiment)
gg <- gg + labs(x="")
gg <- gg + theme_bw()
gg <- gg + theme(strip.background=element_rect(fill="black"))
gg <- gg + theme(strip.text=element_text(color="white", face="bold"))
gg

You could have created another column with the combined variable and value factor instead of using interaction explicitly in the ggplot call, which would also make it easier to change/have more control the x-axis labeling.

enter image description here

Upvotes: 8

Related Questions