help with mosaic plots

Question

[peter.berry5]

I'm trying to plot a mosaic style plot using geom_tile of the common GO terms from a DE experiment. As expected each GO term has multiple DE proteins associated with it. I thought I'd plot the GO ID on the x axis and fill each tile with a protein id. But this results in multiple ids being placed in each tile. How can I have a separate tile for each proteins id?

Enriched_GO_ID <- c("GO:0000062", "GO:0000062", "GO:0000062")
Description <- c("fatty-acyl-CoA binding", "fatty-acyl-CoA binding",
                 "fatty-acyl-CoA binding")
Uniprot.ID <- c("P49748", "P49748", "P22307", "P22307", "P13807", "P13807")
Cell.Line <- c("A", "B", "A", "B", "A", "B")
df <- data.frame(Enriched_GO_ID, Description, Uniprot.ID, Cell.Line)

All <- ggplot(df, aes(x = interaction(Enriched_GO_ID, Description ), 
                          y = Cell.Line))+
  geom_tile(color = "white",
            lwd = 2,
            linetype = 1)+ #now have white border between them
  geom_text(aes(label = Uniprot.ID), color = "white", 
            size = 3)+
  scale_fill_fermenter(palette = "Set2", breaks = c(5,2,1.4,0))+
  coord_fixed()+
  labs(title = "",
      y = "Cell Line Name",
      x = "GO Accession ID")+ 
  theme_classic()+ # white background
  theme(axis.text.x = element_text(size = 15, angle = 90),
        axis.text.y = element_text(size = 15),
        axis.title = element_text(size = 20, face = "bold"))
print(All)
dev.off()

Answer 1

[gglim]

Hi Peter,

This is the output of your code:

And this is what you want? (don't know if it's correct):

Inspired by question here.

Hope this is helpful!

[peter.berry5]

Actually would you mind sharing the code you used? My attempt is below and doesn't give the same result as yours.

df2 <- df %>% 
  group_by(Cell.Line, Enriched_GO_ID) %>%
  mutate(Name1 = paste0(Cell.Line,row_number()))
labs <- with(df2,setNames(as.character(Cell.Line),Name1))
labs2 <- data.frame(labs)

All <- ggplot(df2, aes(x = interaction(Enriched_GO_ID, Description), 
                          y = Name1))+
  geom_tile(color = "white",
            lwd = 2,
            linetype = 1)+ #now have white border between them
  geom_text(aes(label = Uniprot.ID), color = "white", 
            size = 3)+
  scale_fill_fermenter(palette = "Set2", breaks = c(5,2,1.4,0))+
  coord_fixed()+
  scale_y_discrete(labels = df2$Cell.Line)+
  labs(title = "",
      y = "Cell Line Name",
      x = "GO Accession ID")+ 
  theme_classic()+ # white background
  theme(axis.text.x = element_text(size = 15, angle = 90),
        axis.text.y = element_text(size = 15),
        axis.title = element_text(size = 20, face = "bold"))
print(All)

I can't seem to attach the image this generates but the y axis labels in my plot are A, B, A, B, A, B rather than B,B,B,A,A,A. Consequently the protein IDs don't line up with the correct cell line.

[gglim]

Sure thing, here's the code

Enriched_GO_ID <- c("GO:0000062", "GO:0000062", "GO:0000062")
Description <- c("fatty-acyl-CoA binding", "fatty-acyl-CoA binding",
                 "fatty-acyl-CoA binding")
Uniprot.ID <- c("P49748", "P49748", "P22307", "P22307", "P13807", "P13807")
Cell.Line <- c("A", "B", "A", "B", "A", "B")
df <- data.frame(Enriched_GO_ID, Description, Uniprot.ID, Cell.Line)

# Add two lines here
df <- df %>% group_by(Cell.Line) %>% mutate(Name=paste0(Cell.Line,row_number()))
labs <- with(df,setNames(as.character(Cell.Line),Name))

All <- ggplot(df, aes(x = interaction(Enriched_GO_ID, Description ), 
                      y = Name))+ # change the aes y here
  geom_tile(color = "white",
            lwd = 2,
            linetype = 1)+ #now have white border between them
  geom_text(aes(label = Uniprot.ID), color = "white", 
            size = 3)+
  scale_fill_fermenter(palette = "Set2", breaks = c(5,2,1.4,0))+
  coord_fixed()+
  labs(title = "",
       y = "Cell Line Name",
       x = "GO Accession ID")+ 
  theme_classic()+ # white background
  theme(axis.text.x = element_text(size = 15, angle = 90),
        axis.text.y = element_text(size = 15),
        axis.title = element_text(size = 20, face = "bold"))+
  scale_y_discrete(labels=labs) # and a line here

print(All)

Answer 2

[peter.berry5]

dear ggslim,

That wasn't quite what I had in mind but actually I prefer it to what I was looking for.

Thanks

Peter