ordering of labels with bar and point plot - r

Question

I have a bar plot using geom_bar() that I'd like to overlay points using geom_point(). The issues is the ordering of the axis labels. I have 2 groups, group A which I want to show with geom_bar() ordered from high to low and group B which I want to show with points using geom_bar. Group A and B will not always have the same categories but I always want group A shown with bars and ordered from high to low. and
If you run this code you will see just the bar plot correctly ordered. I need the pet supercategory shown first and then the car category. I have defined supercategory as an ordered factor and it is working.
Then within the supercategory, the bars are sorted by gorup A's value form high to low. you can see in the pet category dog is higher than the others and kia is higher than the others in the car category.
library(dplyr)
group = c("A","A","A","B","B","B","A","A","A","B","B","B")
supercategory = c("pet", "pet","pet","pet","pet","pet","car","car","car","car","car","car")
category = c("bird","cat","dog","bird","cat","lizard","ford","chevy","kia","kia","toyota","ford")
supercategory = factor(supercategory, levels= c("pet", "car"), ordered = TRUE)
value=c(3,4,5,4,5,6,1,3,10,8,3,5)
dat = data.frame(group = group,supercategory = supercategory, category = category, value = value )
dat = dat %>% mutate(LABEL = paste0(supercategory, "-",category), HIGH_VALUE = ifelse(group =="A",value,0)) %>%
arrange(supercategory, -HIGH_VALUE)
# after the lines above the data is ordered correctly. first by supercategory then by group A's value from higest to lowest using the HIGH_VALUE field
dat$ROW_NUMBER = 1:nrow(dat)
dat = dat %>% group_by(supercategory,category) %>% mutate(ROW_NUMBER2= min(ROW_NUMBER)) %>% arrange( supercategory ,ROW_NUMBER2)
# after the 2 lines above now the data is sorted by ROW_NUMBER2 which orders the category within supercategory.
# Group A will be be in bars using geom_bar
# group B will be displayed iwht points using geom_point
# The bars and points should be in the order of ROW_NUMBER2
library(ggplot2)
dat$LABEL = factor(dat$LABEL, levels = unique(dat$LABEL), ordered = TRUE)
ggplot(dat[dat$group=="A",] , aes(x = LABEL, y = value))+
geom_bar(stat="identity")
I'd like to keep the ordering of the plot above and just add the points above the bars. And if Group B has a category that is not one of Group A's the point should be to the right of Group A's last bar within whatever supercategory it is in.
But when I try to add the points the ordering gets messed up. Run this code which just adds group B's data as points and you will see the order of the labels gets messed up.
library(ggplot2)
dat$LABEL = factor(dat$LABEL, levels = unique(dat$LABEL), ordered = TRUE)
ggplot(dat[dat$group=="A",] , aes(x = LABEL, y = value))+
geom_bar(stat="identity") +
geom_point(data = dat[dat$group=="B",], aes(x = LABEL, y = value), shape=15, size = 3, color = "blue" )
How can I add this line to the plot:
geom_point(data = dat[dat$group=="B",], aes(x = LABEL, y = value), shape=15, size = 3, color = "blue" )
while keeping group A's ordering?

Answer 1

Each group have not the same set of values, then you have to force the X axis order by adding:
+ scale_x_discrete(limits=dat$LABEL)
Then:
ggplot(data = dat , aes(x = LABEL, y = value) ) +
geom_bar(data = dat[dat$group=="A",], stat="identity") +
geom_point(data = dat[dat$group=="B",], shape=15, size = 3, color = "blue") +
scale_x_discrete(limits=dat$LABEL)

Answer 2

I agree with #Cédric Miachon.
There is a problem of using different x.
A possible way to change the behaviour is to introduce NA's to the not present x:
require(reshape2)
require(dplyr)
require(tidyr)
vector_f <- unique(dat$LABEL)
dat1 <- dat %>%
dcast(group+supercategory~LABEL, value.var = 'value') %>% #casting and gathering
gather(label, value , 3:10)
ggplot() +
geom_bar(data = dat1[dat1$group=="A",],aes(x = factor(label, levels = vector_f), y = value), stat="identity") +
geom_point(data = dat1[dat1$group=="B",], aes(x = factor(label, levels = vector_f), y = value))
##I removed some of the geom_point layout specs