Customizing Plots

ggbulksurv offers several customization options:

  • Subsetting data
  • Colors
  • Legend title and position
  • p-value annotations

This tutorial will go through several most commonly used customizations. To see all possible options, type ?run_bulksurv() in the console.

# Load sample_data (the fictional dataset)
data(sample_data)
dat <- sample_data

The default

For comparison, here’s the default plot:

p <- run_bulksurv(dat,
                  print_stats = FALSE)
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

A customized survival curve

Here is an example of a highly customized survival curve:

p <- run_bulksurv(dat, 
                  sample_order = c("WT", "Drug1", "Drug2"),  # Custom order
                  print_stats = FALSE,                   # Don't print stats
                  palette = c("black", "red", "purple"), # Custom colors
                  legend.title = "",                     # Remove legend title ("Condition")
                  legend.position = c(0.9, 0.9),         # Position legend at top right
                  add.pval = TRUE                       # Add pvalue
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Changing colors

Brewer palettes

RColorBrewer provides a list of useful, aesthetically pleasing palettes that are frequently used in R. To use these palettes, change the palette parameter:

# Using the Set1 brewer palette
p <- run_bulksurv(sample_data, 
             sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             palette = "Set1" # Custom brewer palette
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Journal palettes

ggsci provides a collection of palettes inspired by scientific publishing groups. These colors can also be used in our plots:

# Using the nature palette
p <- run_bulksurv(sample_data, 
             sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             palette = "npg" # Nature palette
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Custom colors (name, hexadecimal)

We can also specify colors, either by name (“black”, “blue”), or by hexadecimal value (“#80080”):

# Using manual colors
p <- run_bulksurv(sample_data, sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             palette = c("black", "red", "#800080") # Custom colors
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Legends

Remove legend title 

p <- run_bulksurv(sample_data, 
                  sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             legend.title = "" # Remove the legend title
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Specify legend position

Specifying a position (right, left, bottom or top) with legend.position:

p <- run_bulksurv(sample_data, 
                  sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             legend.position = "bottom" # Set legend at bottom
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Note that run_bulksurv returns the plot as a ggplot object, so you can further customize it if required:

p + 
  ggplot2::theme(legend.position = "bottom") # Change legend position to be at bottom

Specify legend coordinates

Specific coordinates can also be used. For bottom left, use legend.position = c(0,0); for top right, use legend.position = c(1,1).

p <- run_bulksurv(sample_data, 
                  sample_order = c("WT", "Drug1", "Drug2"),
                  print_stats = FALSE,
                  legend.position = c(0.9, 0.9))
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

P-value and median survival line

To add the median survival line and pvalues, pass the following parameters within run_bulksurv():

  • median survival line: add.median.survival = TRUE"
  • pvalue annotation: add.pval = TRUE

To change the p-value adjustment method, use p_adjust_method.

p <- run_bulksurv(sample_data, 
                  sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             add.median.survival = TRUE, # Add median survival
             p_adjust_method = "bonferroni", # Change padj method
             add.pval = TRUE             # Add pvalue
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition
#> Warning in geom_segment(aes(x = 0, y = max(y2), xend = max(x1), yend = max(y2)), : All aesthetics have length 1, but the data has 3 rows.
#> ℹ Please consider using `annotate()` or provide this layer with data containing
#>   a single row.

Adding confidence intervals 

p <- run_bulksurv(sample_data, sample_order = c("WT", "Drug1", "Drug2"),
             print_stats = FALSE,
             add.conf.int = TRUE  # Add confidence interval
             )
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition

Returning data

By default, run_bulksurv() returns a ggplot2 object for easy plot customization.

If you want to return the pivoted data, summary statistics and the plot object, use returnData = TRUE. This will return all data and plots as a list object.

df_survival <- run_bulksurv(dat, returnData = TRUE)
#> Joining with `by = join_by(x, condition, day, sex, status)`
#> call: formula = Surv(day, status) ~ condition
#> call: formula = Surv(day, status) ~ condition

names(df_survival)
#> [1] "data"            "plot"            "median_survival" "logrank"        
#> [5] "pairwise"

Accessing the individual elements:

head(df_survival$data) # pivoted data
#> # A tibble: 6 × 5
#>       x condition   day sex   status
#>   <int> <fct>     <int> <chr>  <dbl>
#> 1     2 WT            1 M          0
#> 2     2 WT            1 M          1
#> 3     2 WT            1 M          1
#> 4    28 Drug1         1 M          1
#> 5    28 Drug1         1 M          1
#> 6    28 Drug1         1 M          1
head(df_survival$median_survival) # median survival
#> Call: survfit(formula = Surv(day, status) ~ condition, data = df_isurv)
#> 
#>                  n events median 0.95LCL 0.95UCL
#> condition=WT    50     47   20.0      19      21
#> condition=Drug1 50     46    4.5       4       6
#> condition=Drug2 50     42   33.0      32      35
head(df_survival$logrank) # logrank test
#> $n
#> groups
#>    condition=WT condition=Drug1 condition=Drug2 
#>              50              50              50 
#> 
#> $obs
#> [1] 47 46 42
#> 
#> $exp
#> [1] 31.24899 12.48336 91.26765
#> 
#> $var
#>            [,1]      [,2]       [,3]
#> [1,]  19.228227 -4.154711 -15.073516
#> [2,]  -4.154711  8.768527  -4.613816
#> [3,] -15.073516 -4.613816  19.687333
#> 
#> $chisq
#> [1] 186.0854
#> 
#> $pvalue
#> [1] 3.909058e-41
head(df_survival$pairwise) # pairwise logrank with p-value correction
#> $method
#> [1] "Log-Rank test"
#> 
#> $data.name
#> [1] "df_isurv and condition"
#> 
#> $p.value
#>                 WT       Drug1
#> Drug1 2.291558e-14          NA
#> Drug2 3.367080e-24 3.36708e-24
#> 
#> $p.adjust.method
#> [1] "BH"