Question

I want to create stacked bar plot with absolute values for variables in the axis but add the percentages on each bar. This is what my data looks like:

BAM  Mapping    Reads   fraction
bam1   Mapped 22493091 0.88940452
bam1 Unmapped  2796966 0.11059548
bam2   Mapped 27018375 0.88256156
bam3 Unmapped  3595212 0.11743844
bam3   Mapped 27238774 0.89441821
bam4 Unmapped  3215407 0.10558179
bam4   Mapped 19791746 0.82984107
bam4 Unmapped  4058298 0.17015893
bam5   Mapped 23298155 0.83144569
bam5 Unmapped  4723104 0.16855431
bam6   Mapped 22563538 0.83990722
bam6 Unmapped  4300784 0.16009278
bam7   Mapped 23940480 0.88134856
bam7 Unmapped  3222984 0.11865144

I am nearly there (nevermind the x-labels – I’m using long names here):

gp <- ggplot(data=to_graph, aes(x=BAM, y=Reads, fill=Mapping, label=paste(round(fraction*100),"%", sep=""), size = 3,vjust=0, position = "stack")) +
     geom_bar(stat="identity") + 
    geom_text(position="stack")

But there a little niggling square on top of the legend that I want to get rid of. How to do that importantly why does it appear in the 1st place?

Cheers.

Answer 1

That spurious legend appears because you’re putting all sort of stuff inside aes that don’t need to be there. Try this:

ggplot(data=to_graph, aes(x=BAM, y=Reads, fill=Mapping)) +
     geom_bar(stat="identity",position = "stack") + 
    geom_text(aes(label=paste(round(fraction*100),"%", sep="")),size = 3,vjust=0,position="stack")

In general, things go inside aes when you mapping an aesthetic to a variable in your data. If you’re just setting it (i.e. size = 3) that goes outside of aes. Anything you put inside of aes will in general cause ggplot to attempt to create a legend for that aesthetic.

I don’t think I’ve ever seen position mapped inside aes.