I was trying to make a plot showing how many genes with functional annotations are present across my pipeline and connect the dots to show the connection between different boxplot and geom jitter.

But then I got stuck where the lines wouldn't connect to the dots because the position_jitter was somehow different.

The original code was:

tt  %>%
   left_join(phylo %>% select(!NCBI_TaxID), by=c("Genome"="ID")) %>% 
   filter(Order %in% c("Bacillales","Enterobacterales","Flavobacteriales")) %>% 
   ggplot(aes(x=tool,y=n,col=Order,group=Genome))+
   #geom_boxplot()+
   geom_line(position = position_jitter(width = 0.25, seed = 123), col="gray80")+
   geom_point(position = position_jitter(width = 0.25, seed = 123))+
   theme_minimal()+
   facet_wrap(~Order, scales = "free")

turns out the geom_line function sort the data to make sure it goes into one direction. It sorts the data by x, then group and if you have a facet it needs to go first.

Correct code:

tt  %>%
  left_join(phylo %>% select(!NCBI_TaxID), by=c("Genome"="ID")) %>% 
  filter(Order %in% c("Bacillales","Enterobacterales","Flavobacteriales")) %>% 
  mutate(tool=factor(tool, levels=c("All","Patric","Bakta","IMG","Prokka"))) %>% 
  arrange(Order,Genome,tool) %>% 
  ggplot(aes(x=tool,y=n,col=Order))+
  geom_boxplot()+
  geom_line(aes(group=Genome),position = position_jitter(width = 0.25, seed = 123), col="gray80")+
  geom_point(aes(group=Genome),position = position_jitter(width = 0.25, seed = 123))+
  theme_minimal()+
  facet_wrap(~Order, scales = "free")

Final figure:

Error mentionned here