I'm having trouble using ggplot2 to reproduce a violin plot

Question

My data includes a combination of mutations for several genes per sample, the counts, and the MIC (these are discrete) for the antibiotic the bacteria was screened on.

Here is toy data:

structure(list(count = c(4L, 26L, 96L, 93L, 2L, 22L), combo = structure(c(22L, 
19L, 2L, 2L, 20L, 19L), .Label = c("HFALSEAIVTANGGAL", "HFALSEAIVTANGGAP", 
"HFALSEAIVTANGGTP", "HFALSEAIVTANSGAP", "HFALSEAIVTAYGGTP", "HFALSEAIVTTNSGAP", 
"HFALSEAIVTVNGGAL", "HFALSEAIVTVNGGAS", "HFALSEAIVTVNSGAP", "HFALSEAMTTAYGSAA", 
"HFALSEAMTTAYGSAP", "HFALSEAMTTAYGSTP", "HTRUEAIVSANGGAP", "HTRUEAIVTANGGAP", 
"HTRUEAIVTANSGAP", "HTRUEAIVTTNSGAP", "HTRUEAMTTAYGSAP", "YFALSEAIVTANGGAL", 
"YFALSEAIVTANGGAP", "YFALSEAIVTANGGAS", "YFALSEAIVTANSGAP", "YFALSEAIVTTNGGAL", 
"YFALSEAIVTTNGGAP", "YFALSEAIVTTNSGAP", "YFALSEAIVTVNGGAL", "YFALSEAMTTANGGAP", 
"YFALSEAMTTANGGTP", "YFALSEAMTTAYGSAA", "YFALSEAMTTAYGSAP", "YFALSEAMTTAYGSAS", 
"YFALSEAMTTAYGSTP", "YFALSENFNFNFNFNFNFNFNFNFNF", "YTRUEAIVTANGGAL", 
"YTRUEAIVTANGGAP", "YTRUEAMTTAYGSAP"), class = "factor")), row.names = c(NA, 
6L), class = "data.frame")

This goes on and on for each MIC category (there are 8 on a log2 scale).

I am attempting to create a violin plot like this (source: https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(18)30225-1/fulltext ):

I attempted to write some code but didn't get very far:

p <- ggplot(mydata, aes(factor(combo), y=mic)) + 
  geom_violin() +
  geom_boxplot(width=.01, outlier.size=0, fill="grey50") +
  stat_summary(fun.y=median, geom="point", fill="white", shape=21, size=4)
p

Here is what that plot looks like:

I can easily create a count table but I was hoping to have some visual like the image produced below. Can someone steer me in the right direction?

p <- ggplot(mydata, aes(x = combo, y=mic)) 
p + geom_violin()  + geom_jitter(height = 0, width = 0.1, aes(color=factor(combo)))

Answer 1

I couldn't reproduce your plotting code, as it lacks the column mic . However, I think this is what you're looking for:

# load libraries
library(ggplot2)
library(ggforce)

# make toy data
set.seed(1); a <- data.frame(kind = sample(letters[1:10], 500, TRUE), value = sample(1:20, 500, TRUE))

# plot
ggplot(a, aes(x = kind, y = value, colour = kind))+geom_violin()+geom_sina(size = 2.1)

Of course you can play with the format (the shize of the points in the geom_sina call is the most evident).

EDIT

# redefine the first plot, removing the legend:
p1 <- ggplot(a, aes(x = kind, y = value, colour = kind))+
  geom_violin()+
  geom_sina(size = 1.1)+
  theme(legend.position = "none")

# Define toy data for the lower plot:
library(data.table)
set.seed(1)
Genes <- data.table(gene = sample(LETTERS[1:10], 20, TRUE), 
                    n = sample(1:10, 20, TRUE))

# add a coloring variable
Genes[, coloring := cut(n, 3, labels = 1:3)]

# plot the lower plot
p2 <- ggplot(Genes, aes(n, gene, colour = coloring))+
  geom_point(size = 2.8)+
  theme(axis.title = element_blank(), 
        axis.text.x = element_blank(), 
        legend.position = "none", 
        axis.ticks.x = element_blank(), 
        panel.background = element_blank())

# put both plots in the canvas:
library(patchwork)
p1+
 p2+
 plot_layout(ncol = 1, heights = c(.8, .2))

Which produces:

Answer 2

A long time has passed, and now there is an R package called ggupset which makes charts similar to those linked in The Lancet.

https://github.com/const-ae/ggupset

Plot a combination matrix instead of the standard x-axis and create UpSet plots with ggplot2.