ggplot geom_bar: Multiple samples with same name, plotted not side by side

Question

I have a dataframe which is really two separate results merged into one, each gene has two control sample (FRC190 and FRC193) and 3 unknowns. I would like to plot this in the order you read it, with color grouping for each gene.

   SampleID RelativeCopy Loci
1    FRC190     2.040265  ABR
2    FRC193     1.961293  ABR
3     FC124     1.828341  ABR
4    FCP920     2.016274  ABR
5   E-NH021     1.919309  ABR
6    FRC190     1.973149 APRT
7    FRC193     2.027592 APRT
8    FCP604     2.086984 APRT
9    FCP686     2.027592 APRT
10  FCP1130     1.936854 APRT

What I can do is almost that, firstly if I use this code I get the data in the order except the two controls a plotted together.

    df <- within(df, SampleID <- factor(
    df$SampleID, levels = c('FRC190', 'FRC193', 'FCP920', 'E-NH021',
   'FC124', 'FCP1130', 'FCP604', 'FCP686' )))

    ggplot(data = df, aes(x=SampleID,y=RelativeCopy, fill = Loci))+scale_fill_grey() +
    geom_bar(stat="identity", position=position_dodge())+ theme_classic()

The other option I tried was renaming the controls to different names (i.e. FRC190-1 and FRC190-2) then overwriting the names on the axis. The code used for that is

df <- within(df, SampleID <- factor(
df$SampleID, levels = c('FRC190', 'FRC193', 'FCP920', 'E-NH021', 'FC124',"FRC190-2",
 "FRC193-2",'FCP1130', 'FCP604', 'FCP686' )))

ggplot(data = df, aes(x=SampleID,y=RelativeCopy, fill = Loci))+scale_fill_grey()+
geom_bar(stat="identity", position=position_dodge()) + theme_classic() +
scale_x_discrete(breaks=c('FRC190', 'FRC193', 'FCP920', 'E-NH021', 'FC124',"FRC190","FRC193", 'FCP1130', 'FCP604', 'FCP686' ))

This solves the first problem of the grouping of the controls and allows the correct order, but it doesn't allow the use of the same names.

Answer 1

To plot in order you can just use a dummy variable for your x:

p <- ggplot(df, aes(x=1:nrow(df), y=RelativeCopy, fill=Loci)) +
    geom_bar(stat="identity", position=position_dodge()) + 
    theme_classic() + scale_fill_grey()

and then to change the x labels you can use the labels argument for scale_x_*:

p + scale_x_discrete(labels=df$SampleID, breaks=1:nrow(df), limits=1:nrow(df), name='SampleID')

• labels=df$SampleID sets the x tick labels
• breaks=1:nrow(df) says we to plot one tick per bar
• limits=1:nrow(df) if you leave this out the plot just looks a little uncentred (the scale goes up to 11 on the x axis rather than 10 being nrow(df))
• name='SampleID' just sets the label for the x axis.

You may also wish to rotate your axis labels if they run into each other.