5  Customize ER plots

This page demonstrates the options available in plot_er_gof() and how to customize ER plots beyond the built-in options.

See also the function reference for complete documentation.

5.1 Setup

Show the code 
library(tidyverse)
library(BayesERtools)
library(here)

theme_set(theme_bw(base_size = 12))

5.2 Data and model

Show the code 
data(d_sim_binom_cov)

d_sim_binom_cov_2 <-
  d_sim_binom_cov |>
  mutate(
    AUCss_1000 = AUCss / 1000,
    Dose = glue::glue("{Dose_mg} mg")
  )

df_er_ae_hgly2 <- d_sim_binom_cov_2 |> filter(AETYPE == "hgly2")
Show the code 
set.seed(1234)
ermod_bin <- dev_ermod_bin(
  data = df_er_ae_hgly2,
  var_resp = "AEFLAG",
  var_exposure = "AUCss_1000"
)

5.3 plot_er_gof() options

5.3.1 Default plot

Show the code 
plot_er_gof(ermod_bin)
Figure 5.1

5.3.2 Grouping by dose (var_group)

Color data points and add boxplot by dose group:

Show the code 
plot_er_gof(ermod_bin, var_group = "Dose")
Figure 5.2

5.3.3 Show exposure coefficient CI (show_coef_exp)

Display the credible interval for the exposure coefficient:

Show the code 
plot_er_gof(ermod_bin, var_group = "Dose", show_coef_exp = TRUE)
Figure 5.3

Customize the position and size of the coefficient label:

Show the code 
plot_er_gof(
  ermod_bin, 
  var_group = "Dose", 
  show_coef_exp = TRUE,
  coef_pos_x = 2, coef_pos_y = 0.2, coef_size = 3
)
Figure 5.4

5.3.4 Number of bins for observed probability (n_bins)

Change the number of bins used to summarize observed event rates:

Show the code 
plot_er_gof(ermod_bin, var_group = "Dose", n_bins = 6)
Figure 5.5

5.3.5 Confidence interval widths

Control CI widths for observed data summary (qi_width_obs), simulated curve (qi_width_sim), and coefficient (qi_width_coef):

Show the code 
plot_er_gof(
  ermod_bin, 
  var_group = "Dose",
  show_coef_exp = TRUE,
  qi_width_obs = 0.9,
  qi_width_sim = 0.9,
  qi_width_coef = 0.9
)
Figure 5.6

5.3.6 Boxplot options

Control boxplot visibility and appearance:

Show the code 
# Explicitly add boxplot without var_group
plot_er_gof(ermod_bin, add_boxplot = TRUE)
Figure 5.7

Adjust boxplot height and show y-axis title:

Show the code 
plot_er_gof(
  ermod_bin, 
  var_group = "Dose",
  boxplot_height = 0.25,
  show_boxplot_y_title = TRUE
)
Figure 5.8

5.3.7 Caption (show_caption)

The caption is shown by default. Hide it with show_caption = FALSE:

Show the code 
plot_er_gof(ermod_bin, var_group = "Dose", show_caption = FALSE)
Figure 5.9

5.3.8 Using log scale for exposure

Use * instead of + to apply scales to all panels in the combined plot:

Show the code 
plot_er_gof(ermod_bin, var_group = "Dose", show_coef_exp = TRUE) *
  xgxr::xgx_scale_x_log10(guide = guide_axis(minor.ticks = TRUE))
Figure 5.10

5.4 Customize each element of GOF plot

For deeper customization (titles, themes, colors, shapes), use return_components = TRUE to get modifiable ggplot objects, then recombine with combine_er_components().

5.4.1 Basic workflow

Show the code 
# Get components
comps <- plot_er_gof(ermod_bin, var_group = "Dose", return_components = TRUE)

# Modify the main plot
comps$main <- comps$main +
  labs(title = "Exposure-Response Analysis", x = "AUC (ng·h/mL / 1000)")

# Recombine
combine_er_components(comps)
Figure 5.11

5.4.2 Customizing themes, colors, and shapes

Show the code 
comps <- plot_er_gof(ermod_bin, var_group = "Dose", return_components = TRUE)

# Extract original data for custom point layer
origdata <- extract_data(ermod_bin) |>
  mutate(Dose = glue::glue("{Dose_mg} mg"))

comps$main <- comps$main +
  # Add points with custom shapes by dose
  geom_jitter(
    data = origdata,
    aes(x = AUCss_1000, y = AEFLAG, color = Dose, shape = Dose),
    width = 0, height = 0.05, alpha = 0.7
  ) +
  scale_shape_manual(values = c(16, 17, 15)) +
  scale_color_brewer(palette = "Set1", guide = guide_legend(reverse = TRUE)) +
  labs(title = "Custom Theme with Shapes", color = "Dose", shape = "Dose") +
  theme_minimal()
Scale for colour is already present.
Adding another scale for colour, which will replace the existing scale.
Show the code 
comps$boxplot <- comps$boxplot +
  scale_fill_brewer(palette = "Pastel1") +
  scale_color_brewer(palette = "Set1") +
  theme_minimal()

combine_er_components(comps)
Figure 5.12

5.4.3 Adjusting heights and caption

Show the code 
comps <- plot_er_gof(
  ermod_bin, 
  var_group = "Dose", 
  show_caption = TRUE,
  return_components = TRUE
)

comps$main <- comps$main + labs(title = "Adjusted Layout")

# Custom heights and suppress caption
combine_er_components(comps, heights = c(0.7, 0.3), add_caption = FALSE)
Figure 5.13

5.4.4 Applying log scale to components

Show the code 
comps <- plot_er_gof(ermod_bin, var_group = "Dose", return_components = TRUE)

log_scale <- xgxr::xgx_scale_x_log10(guide = guide_axis(minor.ticks = TRUE))

comps$main <- comps$main + log_scale + labs(title = "Log Scale")
comps$boxplot <- comps$boxplot + log_scale

combine_er_components(comps)
Figure 5.14

5.5 Summary

For complete documentation, see the plot_er_gof() reference.

plot_er_gof() options:

Option Description Default
var_group Group data by column NULL
add_boxplot Add exposure boxplot TRUE if var_group set
boxplot_height Relative height of boxplot 0.15
show_boxplot_y_title Show boxplot y-axis title FALSE
n_bins Bins for observed probability 4
qi_width_obs CI width for observed data 0.95
qi_width_sim CI width for model curve 0.95
show_coef_exp Show coefficient CI FALSE
coef_pos_x/y Coefficient label position Auto
coef_size Coefficient label size 4
qi_width_coef CI width for coefficient 0.95
show_caption Show plot caption TRUE
return_components Return components for customization FALSE

Tips:

  • Use * instead of + to apply scales to all panels
  • Use return_components = TRUE for titles, themes, colors, and shapes
  • Components: $main (ER plot), $boxplot, $caption
  • Use combine_er_components() to recombine after modification