CODEWITHSUNDEEP
rtestingggplot2geom-text
DaniCee
DaniCee

Reputation: 3207

R ggplot2: geom_text error (object not found)... aes() clashing?

I just want to add Kruskal-Wallis test results that I compute outside the plot, at the top of some boxplots. I know I could calculate kruskal.test in the same geom_text line, but in this case I'd rather keep it outside since I already have it.

Please check the following MWE:

library(reshape2)
library(ggplot2)
data(iris)
iris$treatment <- rep(c("A","B"), length(iris$Species)/2)
mydf <- melt(iris, measure.vars=names(iris)[1:4])
head(mydf)
mydf$both <- factor(paste(mydf$treatment, mydf$variable, sep=' -- '), levels=(unique(paste(mydf$treatment, mydf$variable, sep=' -- '))))

##Signif levels comparing treatments per Species (regardless of variable)
addkw1 <- as.data.frame(mydf %>% group_by(Species) %>%
                       summarize(p.value = wilcox.test(value ~ treatment)$p.value)) #no need for kruskal, only 2 treatments
##Signif levels comparing variable per Species (regardless of treatment)
addkw2 <- as.data.frame(mydf %>% group_by(Species) %>%
                       summarize(p.value = kruskal.test(value ~ variable)$p.value))
##Signif levels comparing treatment+variable per Species
addkw3 <- as.data.frame(mydf %>% group_by(Species) %>%
                        summarize(p.value = kruskal.test(value ~ both)$p.value))
addkw1$TEST <- "Treatment"
addkw2$TEST <- "Variable"
addkw3$TEST <- "Treat:Var"
addkw <- rbind(addkw1, addkw2, addkw3)
addkw$p.adjust <- p.adjust(addkw$p.value, "BH")
addkw

sp <- "setosa"
addkw0 <- subset(addkw, Species==sp)
df0 <- subset(mydf, Species==sp)

pdf(file="test.df", height=15, width=15)
print(
  ggplot(df0, aes(x=both, y=value, fill=both)) + geom_boxplot() +
    stat_summary(fun.y=mean, geom="point", shape=5, size=4) +
    geom_text(data=addkw0,aes(x=0, y=0, label=p.adjust))
)
dev.off()

As you can see, I just want to plot the boxplots for setosa, for which I have 3 Kruskal-Wallis p-values. I would like 3 lines at the top of the boxplot, like:

KW p-value Treatment = 9.96e01
KW p-value Variable = 1.92e39
KW p-value Treat:Var = 1.18e36

However, I get the following error which I do not know how to solve... my guess is there might be some clashing between the ggplot aes() and the geom_text aes().

Error in FUN(X[[i]], ...) : object 'both' not found

EDIT

Thanks to @baptiste's comment, I managed to avoid the error, but I still struggle at printing the 3 lines with the 3 KW values... Any help?

This is the new code:

png(filename="test.png")
print(
  ggplot(df0, aes(x=both, y=value)) + geom_boxplot(aes(fill=both)) +
  stat_summary(fun.y=mean, geom="point", shape=5, size=4) +
  geom_text(data=addkw0, aes(x=0, y=0, label=paste0("KW pv = ", formatC(p.adjust, format="e", digits=2))), hjust=0)
)
dev.off()

which produces this:

test

Upvotes: 1

Views: 5854

Answers (1)

missuse
missuse

Reputation: 19716

To control the position of geom_text you need to specify the coordinates of each label. Since for all three strings you have specified x = 0 and y = 0 in the geom_text call they overlap.

ggplot(df0, aes(x = both, y = value)) + geom_boxplot(aes(fill = both)) +
  stat_summary(fun.y = mean, geom = "point", shape = 5, size = 4) +
  geom_text(data = addkw0, aes(x = 7, y = c(4.5, 5, 5.5),
                               label = paste0("KW pv = ", formatC(p.adjust, format="e", digits=2))), hjust = 0)

enter image description here

for other applications, like labeling with text on top of each bar, one would need to calculate the appropriate coordinates before hand and incorporate them in a data frame corresponding to addkw0.

Upvotes: 1

Related Questions