How to reorder the place of stacked barplot based on their y value in ggplot?-CodePudding

I'm trying to reorder the stacked bar plot in decreasing order, I mean the bar of Esophageal placed first, but my script doesn't work.

Here is the df

                           Cancer.Study Alteration.Frequency     Alteration.Type Alteration.Count
1                  Esophageal Carcinoma            10.2702703       Amplification               19
2                  Esophageal Carcinoma             1.0810811      Point mutation                2
3        Liver Hepatocellular Carcinoma             0.2652520 Multiple alteration                1
4        Liver Hepatocellular Carcinoma             5.8355438       Amplification               22
5        Liver Hepatocellular Carcinoma             0.7957560      Point mutation                3
6                Stomach Adenocarcinoma             6.4853556       Amplification               31
7          Bladder Urothelial Carcinoma             2.9197080       Amplification               12
8                   Lung Adenocarcinoma             2.7131783       Amplification               14
9          Lung Squamous Cell Carcinoma             2.5948104       Amplification               13
10 Uterine Corpus Endometrial Carcinoma             0.9174312       Amplification                5

And here is my script

ggplot(df,
       aes(fill=factor(Alteration.Type, 
                       levels = c('Point mutation','Amplification','Deep deletion', 'Multiple alteration')),
           x=reorder(Cancer.Study, -Alteration.Frequency)), y=Alteration.Frequency))   
  geom_bar(position="stack", stat="identity") theme_bw()

Thanks for any help

CodePudding user response：

You could use reorder for your fill aesthetic like this:

df <- read.table(text = "                           Cancer.Study Alteration.Frequency     Alteration.Type Alteration.Count
1                  Esophageal.Carcinoma            10.2702703       Amplification               19
2                  Esophageal.Carcinoma             1.0810811      Point.mutation                2
3        Liver.Hepatocellular.Carcinoma             0.2652520 Multiple.alteration                1
4        Liver.Hepatocellular.Carcinoma             5.8355438       Amplification               22
5        Liver.Hepatocellular.Carcinoma             0.7957560      Point.mutation                3
6                Stomach.Adenocarcinoma             6.4853556       Amplification               31
7          Bladder.Urothelial.Carcinoma             2.9197080       Amplification               12
8                   Lung.Adenocarcinoma             2.7131783       Amplification               14
9          Lung.Squamous.Cell.Carcinoma             2.5948104       Amplification               13
10 Uterine.Corpus.Endometrial.Carcinoma             0.9174312       Amplification                5", header = TRUE)

library(ggplot2)
ggplot(df, aes(fill=reorder(Alteration.Type, -Alteration.Frequency),
           x=Cancer.Study, y=Alteration.Frequency))   
  geom_bar(position="stack", stat="identity")   
  labs(fill = "Alteration.Type")  
  theme_bw()  
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))

^{Created on 2022-09-10 with reprex v2.0.2}

Please note: I added some dots to your whitespaces in your data.

CodePudding user response：

By default, reorder uses the respective mean. You however want to order by the absolute count. That's why you have to specify FUN = sum in the reoder() function.

See below with the df table of Quinten

library(tidyverse)
    
ggplot(df, aes(x = reorder(Cancer.Study, -Alteration.Frequency, FUN = sum),
       y = Alteration.Frequency,
       fill = Alteration.Type))   
       geom_bar(position="stack", stat="identity")   
       labs(fill = "Alteration.Type")  
       ylab("Alteration Frequency")  
       xlab("Cancer Study")  
       theme_bw()  
       theme(axis.text.x = element_text(angle = 45, hjust=1))