ggplot interaction plot

interaction_plot.R

library(tidyverse)


timing_results = tribble(
  ~treatment, ~object,    ~mouse_id, ~time, 
  'vehicle',  'familiar', '1',    15,
  'vehicle',  'novel', '1',    55,
  
  'vehicle',  'familiar', '2',    19,
  'vehicle',  'novel', '2',    60,
  
  'vehicle',  'familiar', '3',    25,
  'vehicle',  'novel', '3',    48,
  
  'vehicle',  'familiar', '4',    22,
  'vehicle',  'novel', '4',    59,
  
  'THC',  'familiar', '5',    15,
  'THC',  'novel', '5',    18,
  
  'THC',  'familiar', '6',    19,
  'THC',  'novel', '6',    20,
  
  'THC',  'familiar', '7',    25,
  'THC',  'novel', '7',    26,
  
  'THC',  'familiar', '8',    22,
  'THC',  'novel', '8',    25,
)

timing_results = timing_results %>%
  mutate(x_offset = case_when(
    object == 'familiar' ~ -0.25,
    object == 'novel'    ~  0.25
  ))


#-------------
# Option #1, Use a nudge to adjust the x-axis the desired amount.
# position_nudge technically does not take a vector but this seems to work ok.

ggplot(timing_results, aes(x=treatment, y=time)) +
  geom_line(aes(group=mouse_id), position = position_nudge(y=0, x=timing_results$x_offset)) +
  geom_point(aes(color=object, shape=object), position = position_nudge(y=0, x=timing_results$x_offset), size=5) +
  scale_color_manual(values=c('black','grey70')) +
  scale_shape_manual(values=c(19,15)) +
  theme_classic(15)

#-------------
# Option #2, make the x axis a continuous variable, with labels mapped
# to the treatment names via a factor. This avoids the potential nudge issue not taking vectors.

timing_results = timing_results %>%
  mutate(treatment = as.factor(treatment))

ggplot(timing_results, aes(x=as.numeric(treatment) + x_offset, y=time)) +
  geom_line(aes(group=mouse_id)) +
  geom_point(aes(color=object, shape=object), size=5) +
  scale_x_continuous(breaks = c(1,2), labels=levels(timing_results$treatment)) + 
  scale_color_manual(values=c('black','grey70')) +
  scale_shape_manual(values=c(19,15)) +
  theme_classic(15)