ggplot boxplots with 2 y axes - r

Question

I have been looking everywhere to find out how to ggplot boxplots with 2 y axes.
This is what I want the plot to look like:
boxplot
Example data:
Sample Tumor Score_1 Score_2
1 A 100 -20
2 B 80 -10
3 C 5 -5
4 C 6 -7
5 C 80 -8
6 C 70 -30
7 C 80 -5
8 C 90 -6
9 A 150 -8
10 B 1 -10
11 B 2 -10
12 B 4 -9
13 B 5 -7
14 B 8 -6
15 B 10 -4
16 B 12 -8
17 B 7 -10
18 B 6 -11
19 C 70 -15
20 C 90 -4
21 C 95 -3
22 C 120 -6
23 C 130 -9
24 C 50 -5
25 C 113 -10
26 C 100 -2
27 C 90 -1
28 C 50 -11
29 C 80 -15
30 A 200 -7
31 A 200 -4
32 A 180 -3
33 A 160 -9
34 A 107 -15
35 A 115 -11
36 A 80 -12
37 A 90 -14
38 A 130 -13
39 A 140 -9
40 A 120 -10
myboxplot <- read.csv("Example.csv")
#Set up labels
ylim.prim <- c(0, 500)
ylim.sec <- c(-35, 0)
b <- diff(ylim.prim)/diff(ylim.sec)
a <- b*(ylim.prim[1] - ylim.sec[1])
myboxplot %>%
pivot_longer(cols = c(Score_1, Score_2)) %>%
mutate(name = factor(name, levels = c("Score_1", "Score_2"))) %>%
ggplot(aes(x = Tumor)) +
geom_boxplot(aes(y = value, fill = name)) +
scale_y_continuous(name ="Score 1", sec.axis = sec_axis(~ ((. - a)/b), name = expression("Score 2"))) +
scale_x_discrete(name = "Tumor") +
theme_bw() +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank())+
theme(plot.title = element_text(size = 14, face = "bold"),
text = element_text(size = 12),
#axis.title = element_text(face="bold"),
axis.text.x=element_text(size = 11),
legend.position = "right") +
scale_fill_manual(values = wes_palette("GrandBudapest2"))
I do get the plot in the image (linked above), the problem is my second set of data (the purple boxplots "Score 2") is not being aligned with the second y axis, it is aligning with the first y axis. Since the data is much smaller with a range of -35 to 0, you can't see the difference between the tumor types. Does anyone have any ideas how to change this?
Thank you in advance!

Answer 1

I think the plot you are requesting might be misleading. Instead, how about a facet?
library(tidyverse)
data %>%
pivot_longer(-c("Sample","Tumor"), names_to = "Score") %>%
ggplot(aes( x= Tumor, y = value, fill = Score)) +
geom_boxplot() +
facet_wrap(.~Score, scales = "free")
Or as #NickCox suggests:
data %>%
pivot_longer(-c("Sample","Tumor"), names_to = "Score") %>%
group_by(Score,Tumor) %>%
arrange(value) %>%
mutate(xcoord = seq(-0.25,0.25,length.out = n()),
Tumor = factor(Tumor)) %>%
ggplot(aes( x= Tumor, y = value, fill = Score)) +
geom_boxplot(outlier.shape = NA, coef = 0) +
geom_point(aes(x = xcoord + as.integer(Tumor))) +
facet_wrap(.~Score, scales = "free")

Answer 2

[This was posted when the question was on Cross Validated]
Box plots I find oversold whenever, as usually, there is scope to show more detail. Here is one of several possibilities, a quantile-box plot in Parzen's sense in which for each group a standard box showing median and quartiles is superimposed on a quantile plot, in which the implicit horizontal axis is rank order. The detail that apart from some small integers many values are just multiples of 10 is of interest and should help a little in interpretation.
This plot doesn't use R. People who use R should find doing something similar or better to be trivial -- and those whose favourite software is different should be able to say the same. If not, you need new favourite software.