Add P-values Manually to a ggplot

Add p-values manually to a ggplot, such as box plots, dot plots and stripcharts. Frequently asked questions are available on Datanovia ggpubr FAQ page, for example:

• How to Add P-Values onto Basic GGPlots

• How to Add Adjusted P-values to a Multi-Panel GGPlot

• How to Add P-values to GGPlot Facets

• How to Add P-Values Generated Elsewhere to a GGPlot

• How to Add P-Values onto a Grouped GGPlot using the ggpubr R Package

• How to Create Stacked Bar Plots with Error Bars and P-values

• How to Add P-Values onto Horizontal GGPlots

Usage

stat_pvalue_manual(
  data,
  label = NULL,
  y.position = "y.position",
  xmin = "group1",
  xmax = "group2",
  x = NULL,
  size = 3.88,
  label.size = size,
  p.digits = 3,
  p.format.style = "default",
  p.leading.zero = NULL,
  p.min.threshold = NULL,
  p.decimal.mark = NULL,
  bracket.size = 0.3,
  bracket.nudge.y = 0,
  bracket.shorten = 0,
  color = "black",
  linetype = 1,
  tip.length = 0.03,
  tip.length.ref = c("data", "axis"),
  remove.bracket = FALSE,
  step.increase = 0,
  step.group.by = NULL,
  hide.ns = FALSE,
  vjust = 0,
  coord.flip = FALSE,
  position = "identity",
  inherit.aes = FALSE,
  ...
)

Arguments

data

a data frame containing statistical test results. The expected default format should contain the following columns: group1 | group2 | p | y.position | etc. group1 and group2 are the groups that have been compared. p is the resulting p-value. y.position is the y coordinates of the p-values in the plot.

label

the column containing the label (e.g.: label = "p" or label = "p.adj"), where p is the p-value. Can be also an expression that can be formatted by the glue() package. For example, when specifying label = "t-test, p = {p}", the expression {p} will be replaced by its value. Values inserted through a glue expression are taken from the raw data column and are not rounded by p.digits; to round within a glue expression, wrap the value, e.g. label = "p = \{signif(p, 3)\}" (or {format_p_value(p)} for the p-value house style).

y.position

column containing the coordinates (in data units) to be used for absolute positioning of the label. Default value is "y.position". Can be also a numeric vector.

xmin

column containing the position of the left sides of the brackets. Default value is "group1".

xmax

(optional) column containing the position of the right sides of the brackets. Default value is "group2". If NULL, the p-values are plotted as a simple text.

x

x position of the p-value. Should be used only when you want plot the p-value as text (without brackets).

size, label.size

size of label text.

p.digits

integer indicating the number of digits used to format recognized numeric p-value label columns (label is one of "p", "p.adj", "p.value", "p.val", "pval", "padj"). Default is 3. Set to NULL to print the raw value without rounding (this also disables p.format.style and the related formatting arguments). All other labels are left unchanged: other numeric columns (e.g. "statistic", "n", effect sizes), significance symbols ("p.signif", "p.adj.signif"), already-formatted strings, and glue expressions. A p-named column whose values fall outside [0, 1] is also left as-is. Uses the same formatting engine (format_p_value()) as stat_anova_test() for consistency across layers. The style of this formatting is further controlled by p.format.style, p.leading.zero, p.min.threshold and p.decimal.mark (same arguments as stat_compare_means()).

p.format.style

character string specifying the p-value formatting style applied to recognized numeric p-value label columns (see p.digits). One of: "default" (backward compatible, uses scientific notation), "apa" (APA style, no leading zero), "nejm" (NEJM style), "lancet" (Lancet style), "ama" (AMA style), "graphpad" (GraphPad style), or "scientific" (scientific notation for GWAS). See list_p_format_styles for details. Default is "default", which leaves the rendered p-value labels unchanged.

p.leading.zero

logical indicating whether to include the leading zero before the decimal point (e.g., "0.05" vs ".05"). If provided, overrides the style default.

p.min.threshold

numeric specifying the minimum p-value to display exactly. Values below this threshold are shown as "< threshold" (e.g. p.min.threshold = 0.001 renders very small p-values as "< 0.001"). If provided, overrides the style default.

p.decimal.mark

character string to use as the decimal mark. If NULL, uses getOption("OutDec").

bracket.size

Width of the lines of the bracket.

bracket.nudge.y

Vertical adjustment to nudge brackets by. Useful to move up or move down the bracket. If positive value, brackets will be moved up; if negative value, brackets are moved down.

bracket.shorten

a small numeric value in [0-1] for shortening the with of bracket.

color

text and line color. Can be variable name in the data for coloring by groups.

linetype

linetype. Can be variable name in the data for changing linetype by groups.

tip.length

numeric vector with the fraction that the bracket tips go down to indicate the precise column. Interpreted relative to the reference set by tip.length.ref. Default is 0.03.

tip.length.ref

character string specifying what tip.length is a fraction of. Either "data" (default): fraction of the trained data range (tips scale with the data; existing plots unchanged); or "axis": fraction of the y-axis range (ylim/scale_y_*), giving visually constant tip lengths across plots with different scales (#362).

remove.bracket

logical, if TRUE, brackets are removed from the plot. Considered only in the situation, where comparisons are performed against reference group or against "all".

step.increase

numeric vector with the increase in fraction of total height for every additional comparison to minimize overlap.

step.group.by

a variable name for grouping brackets before adding step.increase. Useful to group bracket by facet panel.

hide.ns

can be logical value or a character vector.

• Case when logical value. If TRUE, hide ns symbol when displaying significance levels. Filter is done by checking the column p.adj.signif, p.signif, p.adj and p.

• Case when character value. Possible values are "p" or "p.adj", for filtering out non significant.

vjust

move the text up or down relative to the bracket. Can be also a column name available in the data.

coord.flip

logical. If TRUE, flip x and y coordinates so that horizontal becomes vertical, and vertical, horizontal. When adding the p-values to a horizontal ggplot (generated using coord_flip()), you need to specify the option coord.flip = TRUE.

position

position adjustment, either as a string, or the result of a call to a position adjustment function.

inherit.aes

If FALSE (the default), overrides the default aesthetics, rather than combining with them. This is most useful for helper functions that define both data and aesthetics and shouldn't inherit behaviour from the default plot specification, e.g. stat_pvalue_manual(). Most layers should use TRUE (the ggplot2 default), but stat_pvalue_manual() uses its own data with different column names than the parent plot data, so FALSE avoids conflicts between parent plot aesthetics (like fill or color) and the p-value annotation data.

...

other arguments passed to the function geom_bracket() or geom_text()

See also

stat_compare_means

Examples

# T-test
stat.test <- compare_means(
  len ~ dose,
  data = ToothGrowth,
  method = "t.test"
)
stat.test
#> # A tibble: 3 × 9
#>   .y.   group1 group2        p   p.adj p.format p.signif method p.format.signif 
#>   <chr> <chr>  <chr>     <dbl>   <dbl> <chr>    <chr>    <chr>  <chr>           
#> 1 len   0.5    1      1.27e- 7 2.5e- 7 1.3e-07  ****     T-test p = 1.3e-07 ****
#> 2 len   0.5    2      4.40e-14 1.3e-13 4.4e-14  ****     T-test p = 4.4e-14 ****
#> 3 len   1      2      1.91e- 5 1.9e- 5 1.9e-05  ****     T-test p = 1.9e-05 ****

# Create a simple box plot
p <- ggboxplot(ToothGrowth, x = "dose", y = "len")
p


# Perform a t-test between groups
stat.test <- compare_means(
  len ~ dose,
  data = ToothGrowth,
  method = "t.test"
)
stat.test
#> # A tibble: 3 × 9
#>   .y.   group1 group2        p   p.adj p.format p.signif method p.format.signif 
#>   <chr> <chr>  <chr>     <dbl>   <dbl> <chr>    <chr>    <chr>  <chr>           
#> 1 len   0.5    1      1.27e- 7 2.5e- 7 1.3e-07  ****     T-test p = 1.3e-07 ****
#> 2 len   0.5    2      4.40e-14 1.3e-13 4.4e-14  ****     T-test p = 4.4e-14 ****
#> 3 len   1      2      1.91e- 5 1.9e- 5 1.9e-05  ****     T-test p = 1.9e-05 ****

# Add p-values manually from stat.test data
# First specify the y.position of each comparison
stat.test <- stat.test %>%
  mutate(y.position = c(29, 35, 39))
p + stat_pvalue_manual(stat.test, label = "p.adj")


# Customize the label with glue expression
# (https://github.com/tidyverse/glue)
p + stat_pvalue_manual(stat.test, label = "p = {p.adj}")



# Grouped bar plots
# %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
ToothGrowth$dose <- as.factor(ToothGrowth$dose)
# Comparisons against reference
stat.test <- compare_means(
  len ~ dose,
  data = ToothGrowth, group.by = "supp",
  method = "t.test", ref.group = "0.5"
)
stat.test
#> # A tibble: 4 × 10
#>   supp  .y.   group1 group2            p       p.adj p.format p.signif method
#>   <fct> <chr> <chr>  <chr>         <dbl>       <dbl> <chr>    <chr>    <chr> 
#> 1 VC    len   0.5    1      0.000000681  0.00000068  6.8e-07  ****     T-test
#> 2 VC    len   0.5    2      0.0000000468 0.000000094 4.7e-08  ****     T-test
#> 3 OJ    len   0.5    1      0.0000878    0.000088    8.8e-05  ****     T-test
#> 4 OJ    len   0.5    2      0.00000132   0.0000026   1.3e-06  ****     T-test
#> # ℹ 1 more variable: p.format.signif <chr>
# Plot
bp <- ggbarplot(ToothGrowth,
  x = "supp", y = "len",
  fill = "dose", palette = "jco",
  add = "mean_sd", add.params = list(group = "dose"),
  position = position_dodge(0.8)
)
bp + stat_pvalue_manual(
  stat.test,
  x = "supp", y.position = 33,
  label = "p.signif",
  position = position_dodge(0.8)
)