Customize the position of `geom

Below is a working example


library(ggplot2)
set.seed(926)
df <- data.frame(expression = rnorm(900),
                 time = c(rnorm(300), rnorm(300, 1, 2), rnorm(300, 2,0.5)),
                 membership = factor(rep(1:3, each = 300)))

ggplot(df, aes(x = time, y = expression, fill = membership))   
  geom_point(shape=21, size = 3)   
  geom_rug(data = subset(df, membership ==3),  sides = "b", color = "green", length = unit(1.5, "cm"))  
  geom_rug(data = subset(df, membership ==2),  sides = "b", color = "blue", length = unit(1, "cm"))  
  geom_rug(data = subset(df, membership ==1),  sides = "b", color = "red")   
  scale_y_continuous(expand = c(0.3, 0))

My hope is something like

Note that I know the options of outside = TRUE, side = "tb" out there. But placing all rug plots at the bottom is what I really hope for.

CodePudding user response：

geom_rug is designed to be drawn at the margins of a plot. It's probably best to use geom_point with a custom symbol in this case:

ggplot(df, aes(x = time, y = expression, fill = membership))   
  geom_point(shape=21, size = 3)   
  geom_point(aes(y = -as.numeric(membership) - 2.5, color = membership), 
             shape = "|", size = 8)  
  geom_hline(yintercept = -3)  
  theme_classic(base_size = 20)  
  scale_y_continuous(breaks = c(-2, 0, 2))

CodePudding user response：

I don't think the position of geom_rug() can be easily customised. I'd recommend to use geom_segment() instead to draw the rugs like you'd want them.

library(ggplot2)
#> Warning: package 'ggplot2' was built under R version 4.2.2
set.seed(926)
df <- data.frame(expression = rnorm(900),
                 time = c(rnorm(300), rnorm(300, 1, 2), rnorm(300, 2,0.5)),
                 membership = factor(rep(1:3, each = 300)))

# Helper variables
limits <- range(df$expression)
step   <- diff(limits) * 0.1
size   <- 0.45 * step

ggplot(df, aes(x = time, y = expression, fill = membership))   
  geom_point(shape=21, size = 3)   
  geom_segment(
    aes(
      colour = membership,
      xend = time,
      y    = limits[1] - as.numeric(membership) * step   size,
      yend = limits[1] - as.numeric(membership) * step - size
    )
  )

^{Created on 2022-12-12 by the reprex package (v2.0.1)}