ggpval allows you to perform statistic tests and add the corresponding p-values to ggplots automatically. P-values can be presented numerically or as stars (e.g. , *). Alternatively, one can also make any text annotation between groups.


# Install `ggpval` from CRAN:

# You can install the lastest ggpval from github with:
# install.packages("devtools")


Simulate data with groups.

A <- rnorm(200, 0, 3)
B <- rnorm(200, 2, 4)
G <- rep(c("G1", "G2"), each = 100)
dt <- data.table(A, B, G)
dt <- melt(dt, id.vars = "G")

A trivial boxplot example

Give the group pairs you want to compare in pairs.

plt <- ggplot(dt, aes(variable, value)) +
  geom_boxplot() +

add_pval(plt, pairs = list(c(1, 2)), fold_change=TRUE)

plot of chunk unnamed-chunk-3

Boxplot with facets

plt <- ggplot(dt, aes(variable, value)) +
  geom_boxplot() +
  geom_jitter() +
add_pval(plt, pairs = list(c(1, 2)))

plot of chunk unnamed-chunk-4

Bar plot

ggpval tries to infer the column which contains the data to do statistical testing. In case this inference was wrong or not possible (for instance the raw data column was not mapped in ggplot object), you can specify the correct column name with response=.

dt[, mu := mean(value),
   by = c("G", "variable")]

dt[, se := sd(value) / .N,
   by = c("G", "variable")]

plt_bar <- ggplot(dt, aes(x=variable, y=mu, fill = variable)) +
  geom_bar(stat = "identity", position = 'dodge') +
  geom_errorbar(aes(ymin=mu-se, ymax=mu+se),
                width = .2) +

add_pval(plt_bar, pairs = list(c(1, 2)), response = 'value')

plot of chunk unnamed-chunk-5

Additional arguments for statistical function can also be directly specified. Here we also the conventional “*” format for significance level.

add_pval(plt_bar, pairs = list(c(1, 2)), 
         test = 't.test',
         alternative = "less",
         response = 'value',
         pval_star = T)

plot of chunk unnamed-chunk-6

Annotate your plot

add_pval(plt, pairs = list(c(1, 2)), annotation = "Awesome")

plot of chunk unnamed-chunk-7

In case you to want give different annotations to each facets, provide your annotation as a list

add_pval(plt, pairs = list(c(1, 2)), annotation = list("Awesome1", "Awesome2"))

plot of chunk unnamed-chunk-8

Bugs and issues

Please report bugs and issues on github issue page: Contributions are welcome.


Thanks to Vicente YĆ©pez for testing and helping with improvement of the package.