Set size line plot with different y axis as addition to a stacked barplot-CodePudding

I would like to plot stacked barplot with added line plot that presents the overall set sizes. I'm plotting stacked barplot in ggplot2 without problems however additional line with different y axis is the difficulty. I'm using long-formated table as input, so there is no 'overall size' column.

Code to reproduce sample table:

df <- data.frame(Sample=c("S1","S2","S3","S4","S5","S6"), A=c(30,52,50,81,23,48), B=c(12,20,15,22,30,14), C=c(rep(15,6)))
df.melt <- melt(setDT(df), id.vars = "Sample", variable.name = "Group")

Head of the table:

   Sample Group value
1:     S1     A    30
2:     S2     A    52
3:     S3     A    50
4:     S4     A    81
5:     S5     A    23
6:     S6     A    48

Code to draw stacked barplot:

ggplot(df.melt, aes(x = Sample, y = value, fill = Group))  
  geom_col(position = position_fill(reverse = TRUE))  
  theme(axis.text.x=element_text(angle=45, hjust=1), legend.title=element_blank())  
  scale_fill_brewer(palette="Set3")  
  ylab("% of Total")  
  scale_y_continuous(labels = percent)  
  scale_x_discrete(limits  = unique(df.melt$Sample))

Therefore the line would run through six stacked bars pointing the size of each set i.e. for sample S1 it would be 57 (A B C), and y axis labels to the right of the plot would show set size range.

CodePudding user response：

You can put the data set directly in the geom. This allows you to use different data sets for each geom. Secondary axis are a bit tricky. They need to be a function of the primary axis and the data adjusted accordingly. I've used 120 as the adjustment factor.


percent <- c("0%", "25%", "50%", "75%", "100%")

set_sizes <- df %>%
    rowwise %>%
    mutate(Size = sum(A, B, C))
    
    
ggplot()  
    geom_col(df.melt, mapping = aes(x = Sample, y = value, fill = Group),position = position_fill(reverse = TRUE))  
    geom_line(set_sizes, mapping = aes(x = Sample, y = Size / 120, group = 1))  
    scale_y_continuous(name = "% of Total", labels = percent, sec.axis = sec_axis(~ .*120, name  = "Sample Size"))              
    theme(axis.text.x=element_text(angle=45, hjust=1), legend.title=element_blank())  
    scale_fill_brewer(palette="Set3")  
    scale_x_discrete(limits  = unique(df.melt$Sample))

CodePudding user response：

Alternatively, you can use cowplot to arrange two independent plots on top of each other, e.g.:

suppressMessages(invisible(lapply(c("data.table", "ggplot2", "cowplot"), 
    require, character.only=TRUE)))
df <- data.table(Sample=c("S1","S2","S3","S4","S5","S6"), 
    A=c(30,52,50,81,23,48), B=c(12,20,15,22,30,14), C=c(rep(15,6)))
df.melt <- melt(df, id.vars = "Sample", variable.name = "Group")
percent <- paste0(sprintf("%s", seq(0, 100, 25)), "%")
p1 <- ggplot(df.melt, aes(x = Sample, y = value, fill = Group))  
  geom_col(position = position_fill(reverse = TRUE))  
  theme(axis.text.x=element_text(angle=45, hjust=1), legend.title=element_blank())  
  scale_fill_brewer(palette="Set3")  
  ylab("% of Total")  
  scale_y_continuous(labels = percent)  
  scale_x_discrete(limits  = unique(df.melt$Sample))

p2 <- ggplot(df.melt[, .(value=sum(value)), by="Sample"], 
    aes(x = Sample, y = value, group=1))  
    geom_line()   
    scale_x_discrete(labels = NULL, breaks = NULL)   
    labs(x = NULL)
plot_grid(p2, NULL, p1, align="hv", nrow=3, axis='tlbr', rel_heights=c(1, -.28, 4), greedy=FALSE)

^{Created on 2022-02-20 by the reprex package (v2.0.1)}