Introduction

Tukey Pos-Hoc Test is an statistical test to avaliate which elements of a ANOVA (Analysis of Variation) analysis had variate between themself. As another analysis, Tukey Pos-Hoc Test can be plotted on graphics. On R Programing language, Tukey Pos-Hoc Test can be easily plotted with the plot() function.

Plotting with plot()

sp1<-c(22,34,26,24,30)
sp2<-c(35,43,44,29,38)
sp3<-c(45,54,39,55,49)
sp4<-c(33,35,36,37,42)
dados<-data.frame(sp1,sp2,sp3,sp4)
dat<-stack(dados)
anova = aov(dat$values~dat$ind)
tk_teste <- TukeyHSD(anova)
plot(tk_teste, col="black", cex.axis=0.75)

As it can be understood, the bars that do not cruse the vertical line means a variation between two elements. In our situation, sp1-sp2 element means a variation between sp1 and sp2 groups. Although, it is a limitated plot graphic, in some situations. To resolute that, i created a function to make a ggplot plot Tukey Pos-Hoc Test.

Getting data

Loading the data

To make our analysis, we loading a databse of four landsites, which one containing data about the herpetofauna species diversity on number of species per landsites, assigning to a database called testeanv.

setwd("G:/Meu Drive/R")
library(readxl)
testeanv<-read_xlsx("testeanova.exemplo.xlsx")
testeanv

## # A tibble: 20 × 2
##    Valores Locais
##      <dbl> <chr> 
##  1      22 L1    
##  2      34 L1    
##  3      26 L1    
##  4      24 L1    
##  5      30 L1    
##  6      35 L2    
##  7      43 L2    
##  8      44 L2    
##  9      29 L2    
## 10      38 L2    
## 11      45 L3    
## 12      54 L3    
## 13      39 L3    
## 14      55 L3    
## 15      49 L3    
## 16      33 L4    
## 17      35 L4    
## 18      36 L4    
## 19      37 L4    
## 20      42 L4

R required database formate

To make the ANOVA analysis on R, it is necessary to formate the database in one way: a single colune with all the data of herpetofauna species diversity and an another single colune with the all landsites, corresponding with the previous rows data, as we can see above on the loaded database. To optimaze our analysis, it is necessary to certify if our data are on the R required database formate to do an ANOVA analysis.

Doing the analysis

ANOVA test

Our first step here is to make an ANOVA analysis with the testeanv database, therefore creating a aov objet.

ANOVA<-aov(testeanv$Valores~testeanv$Locais)
summary(ANOVA)

##                 Df Sum Sq Mean Sq F value   Pr(>F)    
## testeanv$Locais  3   1129   376.3   12.98 0.000149 ***
## Residuals       16    464    29.0                     
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

It was did an ANOVA analysis, with the command aov(), assigning it to a objet called ANOVA. Next step, we used the summary() command on ANOVA objet. That result had show us: p=0.000149. It means the groups variate themself, so it is available to proceed to the Tukey Pos-Hoc Test.

Tukey test

With a ANOVA object, it can do a the Tukey Pos-Hoc Test, using TukeyHSD() command, assigning to an objetc called TUKEY.

TUKEY<-TukeyHSD(ANOVA)
TUKEY

##   Tukey multiple comparisons of means
##     95% family-wise confidence level
## 
## Fit: aov(formula = testeanv$Valores ~ testeanv$Locais)
## 
## $`testeanv$Locais`
##        diff         lwr       upr     p adj
## L2-L1  10.6   0.8557176 20.344282 0.0306731
## L3-L1  21.2  11.4557176 30.944282 0.0000653
## L4-L1   9.4  -0.3442824 19.144282 0.0606123
## L3-L2  10.6   0.8557176 20.344282 0.0306731
## L4-L2  -1.2 -10.9442824  8.544282 0.9844430
## L4-L3 -11.8 -21.5442824 -2.055718 0.0151889

The Tukey test results showd sites 3-1, sites 3-2 and sites 4-3 differ significatilly themself. Then, we do the ggplot Tukey Pos-Hoc Test plotting.

Created function

Creating the function

GGTukey<-function(Tukey){
  A<-require("tidyverse")
  if(A==TRUE){
    library(tidyverse)
  } else {
    install.packages("tidyverse")
    library(tidyverse)
  }
  B<-as.data.frame(Tukey[1])
  colnames(B)[2:3]<-c("min",
                      "max")
  C<-data.frame(id=row.names(B),
                min=B$min,
                max=B$max)
  D<-C%>%
    ggplot(aes(id))+
    geom_errorbar(aes(ymin=min,
                     ymax=max),
                 width = 0.2)+
    geom_hline(yintercept=0,
               color="red")+
    labs(x=NULL)+
    coord_flip()+
    theme(text=element_text(family="TimesNewRoman"),
            title=element_text(color="black",size=15),
            axis.text = element_text(color="black",size=10),
            axis.title = element_text(color="black",size=10),
            panel.grid=element_line(color="grey75"),
            axis.line=element_blank(),
            plot.background=element_rect(fill="white",color="white"),
            panel.background=element_rect(fill="white"),
            panel.border = element_rect(colour = "black", fill = NA,size=0.59),
            legend.key= element_rect(color="white",fill="white")
          )
  return(D)
}

Understandding the function and their elements

Starting the 1st commands and check the required packages

To make that function, it was used the function(){} command, when the tukey objetc was the condicional object. In that command, to create a ggplot plotting and get manipulatting the data, we get the package called tidyverse: it load packages, like the dplyr and ggplot2 packages, used to make our plots and get manipulatting our data. One possible problema would be a no dowloaded-tidyverse decive. To resolve that situation, it was:

Used the require() command. The require() ckeck out the downloaded used RStudio to average if an object is existent or package had been downloaded. If the command average the condicional is true, it will print an TRUE message. Then, that command had been assigninged to an A object.
Using the created A object, it was creat the condicional commands if(){} and else{}. If the A objet value equals to a TRUE print message, the RStudio only will load the Tidyverse package, as can be realised in the if(){} condicional command. But if the the A objet value equals to a FALSE print message, the RStudio will start to install the tidyverse package, to after load that package.

That commands were created to work with two possible dinstictic situations:

The user had previously downloaded the tidyverse package, when it was necessary only load the package;
The user do not had previously downloaded the tidyverse package, when it was necessary firs install it, to posteriorly load that package.

Creating the required data frames

Our next step is to change the Tukey object in a passively dataframe one, to make a ggplot plot. To do this, the only column of the Tukey object, containg the required data, was changed into a data frame object, with the as.data.frame() command, and assigning to the B object. The B object had been created, to padronize our plots, it was used the colnames() command, assigning to the 2nd to the 3rt column the names lwr and upr, respectivelly.

So, the 1st data frame object was created. Our next step is to make a 2nd data frame object. The 2nd data frame object will be the one who makes the ggplot plots. To make that 2nd data frame, we use the data.frame() command, using the 1st data frame data, following the steps:

Creating a identify column: that column will be used to show how analysis are been represeted by the lines, across the axis id. To make it, it was used the row.names() command, using B object, our 1st data frame, transforming it into a new data frame column, called id.
Creating a minimal values columns: that column will be used to the minimal plot lines values. To make it, it was assigned the lwr column B data frame to a new column, called min.
Creatung a maximal values columns: that column will be used to the maximal plot lines values. To make it, it was assigned the upr column B data frame to a new column, called max.

All of them was used to creat, with the data.frame() command, our 2nd data frame. called C. That new data frame object will be used to creat the ggplot plot.

Making the ggplot plot command

Using the data frame called C, containg the require data, we used the ggplot() command, using the column id on aes() ggplot function. To creat the lines, it is used the geom_errorbar() ggplot command, with min column values being the Y minnimal limits and max column values being the X maximal limits. After that, using the geom_hline() ggplot command to creat a red line interceptting the 0 Y axis value, matching who variation is significant (p<0.05). To finish the main plot layer, is used the coord_flip() ggplot command, to reverse the plot axis. The final ggplot command is the theme(), setting the aestthetical plot layer. The ggplot() command was assigning to a object called D.

Finishing the condictional function command

To finish the function, was used a return() command, to the D object. That command do the function, after made, return all the coded command in this function. The function is made.

Ggplot tukey test plotting

Creating a ggplot tukey test plotting

GGTukey(TUKEY)

## Carregando pacotes exigidos: tidyverse

## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.1 ──

## ✔ ggplot2 3.3.6     ✔ purrr   0.3.4
## ✔ tibble  3.1.6     ✔ dplyr   1.0.9
## ✔ tidyr   1.2.0     ✔ stringr 1.4.0
## ✔ readr   2.1.2     ✔ forcats 0.5.1

## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()

## Warning in grid.Call(C_stringMetric, as.graphicsAnnot(x$label)): font family not
## found in Windows font database

Differing by color

Another way to make that ggplot graphic is differing the elements into significant and not significant by colors, insted using a vertical line crusing the 0 X axis value. That configuration allows identify the significant variantions using visual signals, being more easily interpretabel.

Making the function

GGTukey.2<-function(Tukey){
  A<-require("tidyverse")
  if(A==TRUE){
    library(tidyverse)
  } else {
    install.packages("tidyverse")
    library(tidyverse)
  }
  B<-as.data.frame(Tukey[1])
  colnames(B)[2:4]<-c("min",
                      "max",
                      "p")
  C<-data.frame(id=row.names(B),
                min=B$min,
                max=B$max,
                idt=ifelse(B$p<0.05,
                           "significant",
                           "not significant")
                )
  D<-C%>%
    ggplot(aes(id,color=idt))+
    geom_errorbar(aes(ymin=min,
                      ymax=max),
                  width = 0.5,
                  size=1.25)+
    labs(x=NULL,
         color=NULL)+
    scale_color_manual(values=c("red",
                                "green")
                       )+
    coord_flip()+
    theme(text=element_text(family="TimesNewRoman"),
            title=element_text(color="black",size=15),
            axis.text = element_text(color="black",size=10),
            axis.title = element_text(color="black",size=10),
            panel.grid=element_line(color="grey75"),
            axis.line=element_blank(),
            plot.background=element_rect(fill="white",color="white"),
            panel.background=element_rect(fill="white"),
            panel.border = element_rect(colour = "black", fill = NA,size=0.59),
            legend.key= element_rect(color="white",fill="white")
          )
  return(D)
}

In that script, the only four difference is:

name the fourth colunm on the B data frame object.
creat a new colunm in the C data frame, named idt.
in aes() command, was assigned color=idt.
to define the color by the significant or not significant elements, based on p-value, using scale_color_manual() command.

Creating a ggplot tukey test plotting

GGTukey.2(TUKEY)

## Warning in grid.Call(C_stringMetric, as.graphicsAnnot(x$label)): font family not
## found in Windows font database

## Warning in grid.Call(C_textBounds, as.graphicsAnnot(x$label), x$x, x$y, : font
## family not found in Windows font database

## Warning in grid.Call(C_textBounds, as.graphicsAnnot(x$label), x$x, x$y, : font
## family not found in Windows font database

Using the both ways

Making the function

GGTukey.3<-function(Tukey, Style){
  GT<-function(Tukey){
    A<-require("tidyverse")
    if(A==TRUE){
      library(tidyverse)
    } else {
      install.packages("tidyverse")
      library(tidyverse)
    }
    B<-as.data.frame(Tukey[1])
    colnames(B)[2:3]<-c("min",
                        "max")
    C<-data.frame(id=row.names(B),
                  min=B$min,
                  max=B$max)
    D<-C%>%
      ggplot(aes(id))+
      geom_errorbar(aes(ymin=min,
                        ymax=max),
                    width = 0.2)+
      geom_hline(yintercept=0,
                 color="red")+
      labs(x=NULL)+
      coord_flip()+
      theme(text=element_text(family="TimesNewRoman"),
            title=element_text(color="black",size=15),
            axis.text = element_text(color="black",size=10),
            axis.title = element_text(color="black",size=10),
            panel.grid=element_line(color="grey75"),
            axis.line=element_blank(),
            plot.background=element_rect(fill="white",color="white"),
            panel.background=element_rect(fill="white"),
            panel.border = element_rect(colour = "black", fill = NA,size=0.59),
            legend.key= element_rect(color="white",fill="white")
            )
    return(D)
  }
  GT2<-function(Tukey){
    A<-require("tidyverse")
    if(A==TRUE){
      library(tidyverse)
    } else {
      install.packages("tidyverse")
      library(tidyverse)
    }
    B<-as.data.frame(Tukey[1])
    colnames(B)[2:4]<-c("min",
                        "max",
                        "p")
    C<-data.frame(id=row.names(B),
                  min=B$min,
                  max=B$max,
                  idt=ifelse(B$p<0.05,
                             "significant",
                             "not significant")
    )
    D<-C%>%
      ggplot(aes(id,color=idt))+
      geom_errorbar(aes(ymin=min,
                        ymax=max),
                    width = 0.5,
                    size=1.25)+
      labs(x=NULL,
           color=NULL)+
      scale_color_manual(values=c("red",
                                  "green")
      )+
      coord_flip()+
      theme(text=element_text(family="TimesNewRoman"),
            title=element_text(color="black",size=15),
            axis.text = element_text(color="black",size=10),
            axis.title = element_text(color="black",size=10),
            panel.grid=element_line(color="grey75"),
            axis.line=element_blank(),
            plot.background=element_rect(fill="white",color="white"),
            panel.background=element_rect(fill="white"),
            panel.border = element_rect(colour = "black", fill = NA,size=0.59),
            legend.key= element_rect(color="white",fill="white")
            )
    return(D)
  }
  GR<-if(Style==1){
    return(GT(Tukey))
  } else if(Style==2){
    return(GT2(Tukey))
  }
  return(GR)
}

In this script combining the previously both functions. To chose which plot aesthectic, more than TukeyHSD object, was created a variable named Style. To make the algoritimical process, it was used the if() and else if() condictional command, in which Style=1 creat the ggplot with a vertical red line, and Style=2 Creat the ggplot using collors to differing the significan and not significant elements.

Creating the plots

GGTukey.3(TUKEY,Style=1)

## Warning in grid.Call(C_textBounds, as.graphicsAnnot(x$label), x$x, x$y, : font
## family not found in Windows font database

GGTukey.3(TUKEY,Style=2)

## Warning in grid.Call(C_textBounds, as.graphicsAnnot(x$label), x$x, x$y, : font
## family not found in Windows font database

## Warning in grid.Call(C_textBounds, as.graphicsAnnot(x$label), x$x, x$y, : font
## family not found in Windows font database

## Warning in grid.Call(C_textBounds, as.graphicsAnnot(x$label), x$x, x$y, : font
## family not found in Windows font database

Plotting Tukey Test on ggplot

Edson Nilton de Moura Silva Júnior

19/03/2022

Introduction

Plotting with plot()

Getting data

Loading the data

To make our analysis, we loading a databse of four landsites, which one containing data about the herpetofauna species diversity on number of species per landsites, assigning to a database called testeanv.

R required database formate

Doing the analysis

ANOVA test

Our first step here is to make an ANOVA analysis with the testeanv database, therefore creating a aov objet.

It was did an ANOVA analysis, with the command aov(), assigning it to a objet called ANOVA. Next step, we used the summary() command on ANOVA objet. That result had show us: p=0.000149. It means the groups variate themself, so it is available to proceed to the Tukey Pos-Hoc Test.

Tukey test

With a ANOVA object, it can do a the Tukey Pos-Hoc Test, using TukeyHSD() command, assigning to an objetc called TUKEY.

The Tukey test results showd sites 3-1, sites 3-2 and sites 4-3 differ significatilly themself. Then, we do the ggplot Tukey Pos-Hoc Test plotting.

Created function

Creating the function

Understandding the function and their elements

Starting the 1st commands and check the required packages

Used the require() command. The require() ckeck out the downloaded used RStudio to average if an object is existent or package had been downloaded. If the command average the condicional is true, it will print an TRUE message. Then, that command had been assigninged to an A object.

That commands were created to work with two possible dinstictic situations:

The user had previously downloaded the tidyverse package, when it was necessary only load the package;

The user do not had previously downloaded the tidyverse package, when it was necessary firs install it, to posteriorly load that package.

Creating the required data frames

So, the 1st data frame object was created. Our next step is to make a 2nd data frame object. The 2nd data frame object will be the one who makes the ggplot plots. To make that 2nd data frame, we use the data.frame() command, using the 1st data frame data, following the steps:

Creating a identify column: that column will be used to show how analysis are been represeted by the lines, across the axis id. To make it, it was used the row.names() command, using B object, our 1st data frame, transforming it into a new data frame column, called id.

Creating a minimal values columns: that column will be used to the minimal plot lines values. To make it, it was assigned the lwr column B data frame to a new column, called min.

Creatung a maximal values columns: that column will be used to the maximal plot lines values. To make it, it was assigned the upr column B data frame to a new column, called max.

All of them was used to creat, with the data.frame() command, our 2nd data frame. called C. That new data frame object will be used to creat the ggplot plot.

Making the ggplot plot command

Finishing the condictional function command

To finish the function, was used a return() command, to the D object. That command do the function, after made, return all the coded command in this function. The function is made.

Ggplot tukey test plotting

Creating a ggplot tukey test plotting

Differing by color

Another way to make that ggplot graphic is differing the elements into significant and not significant by colors, insted using a vertical line crusing the 0 X axis value. That configuration allows identify the significant variantions using visual signals, being more easily interpretabel.

Making the function

In that script, the only four difference is:

name the fourth colunm on the B data frame object.

creat a new colunm in the C data frame, named idt.

in aes() command, was assigned color=idt.

to define the color by the significant or not significant elements, based on p-value, using scale_color_manual() command.

Creating a ggplot tukey test plotting

Using the both ways

Making the function

Creating the plots