PPS 3-State Model

by Nicholas Hamilton 16/12/2013 | 4:32 19 Posted in Case Studies, Examples

I recently found THIS site on the internet which I thought would be another good case-study.

As a standalone figure, it can be rendered as follows with ggtern:

#Load the data
df <- read.table(file = "~/Desktop/PPS-data.txt", header = T)

#Create the plot and store
plot <- ggtern(data = df, aes(x = Xyp, y = XO, z = XY)) + 
               geom_point(aes(fill = Root),
                          size = 4, 
                          shape = 21, 
                          color = "black") + 
               ggtitle("PPS 3-State Model") + 
               labs(fill = "Root States") + 
               theme_tern_rgbw() + 
               theme(legend.position      = c(0,1), 
                     legend.justification = c(0, 1))

#Render
plot

#Load the data

df <- read.table(file = "~/Desktop/PPS-data.txt", header = T)

#Create the plot and store

plot <- ggtern(data = df, aes(x = Xyp, y = XO, z = XY)) +

geom_point(aes(fill = Root),

size = 4,

shape = 21,

color = "black") +

ggtitle("PPS 3-State Model") +

labs(fill = "Root States") +

theme_tern_rgbw() +

theme(legend.position = c(0,1),

legend.justification = c(0, 1))

#Render

plot

However, in order to distinguish the different categories, perhaps a facetted approach would be better:

#Render w/ Facetting
plot + facet_wrap(~Root)

1 2	#Render w/ Facetting plot + facet_wrap(~Root)

Tags:categorical, categories, diagrams, facet, facetting, model, plot, R, ternary

Nick Hamilton

Author of the ggtern package. PhD candidate at the University of New South Wales in the School of Materials Science and Engineering. My field is in the simulations of metallic glasses via several advanced simulation techniques such as QMD and RMC.

Comments

Hammao 16/05/2014 | 8:48

Can you attach the txt file please, so a beginner like me can learn.

Reply
Chris 24/07/2015 | 5:26

Very cool, but as HAMMAO notes, could you make sure the data are available for these examples so we can replicate them?

Reply
Nicholas Hamilton 25/07/2015 | 1:19

Sorry, I was travelling at the time and didn’t save the data down. The original source file that I was referring to is no longer on the web. The data is fairly trivial, surely you can come up with an equivalent set.

Reply

Janet 3/09/2015 | 5:13

Hello, Nick,
Could you please explain what does “root” mean, and how can we use roots? Thanks!

Reply

Nicholas Hamilton 3/09/2015 | 6:00

Janet, I think this example is becoming too problematic, because I have lost the original data. Root, was one of the named columns in the source dataset. By using the term ‘Root”, mapped to fill and also the faceting, this means that for every unique value of ‘Root’ (in this case there were 4 unique values) a new facet gets generated, and also a new point fill colour. Hope this helps. Cheers, N

Reply

Janet 3/09/2015 | 7:08

Thank you Nick!
I want to draw a triangle plot for some metabolite data. So each subject has three kinds of metabolites, which is three-way in the triangle plot, and subjects are from two city. I want to use different colors to represent subjects from two cities.
I originally used triangle.plot() to draw it, but it does not support color adjustment (does it?)
Now I wish to use this package. But somehow it doesnt work ( I am a beginner…)
So my data is like
id city a% b% c%
1 1 10 10 80
2 1 5 5 90
3 2 7 7 86

Could you please give me some instructions based on my situation?

Thanks a lot!

Reply

Nicholas Hamilton 3/09/2015 | 9:48

Janet,

I have emailed you, for anyone else, see below:

Probably get rid of the special characters (% signs) in the titles of your data-frame and save down somewhere (my case desktop) as ‘sample.txt’

id city a b c
1 1 10 10 80
2 1 5 5 90
3 2 7 7 86

id city a b c

1 1 10 10 80

2 1 5 5 90

3 2 7 7 86

Now you can load the data, and create a base plot:

library(ggtern)
df <- read.table("~/Desktop/sample.txt",
                 header=T)

#Create Base Plot
base <- ggtern(data=df,aes(a,b,c,
                       colour=factor(city))) + 
  geom_point(size=5) + 
  theme_rgbw() + 
  custom_percent("%") +
  labs(colour="City") + 
  theme(legend.position=c(0,1),
        legend.justification=c(0,1))
print(base)

library(ggtern)

df <- read.table("~/Desktop/sample.txt",

header=T)

#Create Base Plot

base <- ggtern(data=df,aes(a,b,c,

colour=factor(city))) +

geom_point(size=5) +

theme_rgbw() +

custom_percent("%") +

labs(colour="City") +

theme(legend.position=c(0,1),

legend.justification=c(0,1))

print(base)

Because the data is in the lower right corner, use limit_tern() to focus

#Draw same plot above using limiting region
lim <- base + 
       limit_tern(.3,.3,1)
print(lim)

#Draw same plot above using limiting region

lim <- base +

limit_tern(.3,.3,1)

print(lim)

Heath Blackmon 5/09/2015 | 9:57

Hi all,
This was my dataset originally. The original data (observed or “emp”) comes from the type of sex chromosomes that we observe in living beetles. This provides a single point on the plot. All of the other points are the results of simulating the evolution of sex chromosomes in beetles. Part of the process in doing the simulation is choosing a state for the root or beginning of the simulation. I have posted an update where you can simulate data to use for these type of plots under a phylogenetic (evolutionary) or non-phylogenetic model.

new post: http://coleoguy.blogspot.com/2015/09/ternary-diagram-working-example.html

Here is a link to the paper that I was working on where you can see the actual application of the approach:
http://www.genetics.org/content/197/2/561.short

Reply

Nicholas Hamilton 5/09/2015 | 10:18

Heath,

Thankyou so much for getting in touch. Are you happy for this case study to remain, would you like me to update/modify it in any way?

Cheers,

Nick
Heath Blackmon 6/09/2015 | 3:47

Nick,
No it looks great I just wanted to provide a bit of background and links to the updated post.
cheers
Heath

Kinjal 3/06/2016 | 7:46

0.20
1.76
98.04

2.67
10.67
86.67

8.57
20.00
71.43

14.55
21.82
63.64

40.00
40.00
20.00

50.00
33.33
16.67

58.33
25.00
16.67

69.57
17.39
13.04

81.82
9.09
9.09

Hi.. My dataset for the ternary plot are as above.. I tried to load the dataset in R software but I could not do it using load workspace from file menu.. can u please help on this by giving the screenshots or videos for how to use R software-ggtern??

thanks,

kinjal parikh

kinjalparikh1@yahoo.com

research scholar

faculty of pharmacy

the m s university of baroda

gujarat, india

Reply
Kinjal 3/06/2016 | 7:48

consider it as a b c. a b c. idk how it came into 1 column instead of 3..

Reply

Nicholas Hamilton 13/08/2016 | 10:04

Kinjal, try this:

Example for Kinjal
R

df = read.table("kinjal.csv") ix = seq(1,nrow(df),by=3) df = data.frame(A=df[ix,],B=df[ix+1,],C=df[ix+2,]) ggtern(df,aes(A,B,C)) + geom_point()

1
2
3
4

df = read.table("kinjal.csv")
ix = seq(1,nrow(df),by=3)
df = data.frame(A=df[ix,],B=df[ix+1,],C=df[ix+2,])
ggtern(df,aes(A,B,C)) + geom_point()

Reply

Linda 31/01/2017 | 11:26

Hi there,

how do I plot points that lie directly at one of the axes so that they are fully visible and not cut because they partially extend out of the plotting region? The point close to the C vertex in the plot above (from 13/08/2016 | 10:04) is an example of this undesirable visualization.

I would be grateful for any hints on how to fix this. Thanks!

Linda

Reply

Nicholas Hamilton 31/01/2017 | 11:41
Linda, thanks for you question. There are a couple of ways:
1. Turn the clipping mask off using theme_nomask(), or
2. Put the mask geometry explicitly underneath the points layer, ie geom_mask(...) + geom_point(...).
If the mask layer is not put somewhere explicitly, it is added as the final layer, unless the mask is switched off entirely.

Reply

Linda 31/01/2017 | 12:02

Many thanks for your reply! Works perfectly now!

Reply
dinesh 17/11/2017 | 4:50

Hello dear Hamilton

How we can add geometric mean in ggtern package to ternary plot

Reply

86Elden 21/08/2017 | 5:42

I have noticed you don’t monetize your site, don’t waste your traffic, you can earn extra
cash every month because you’ve got hi quality content.
If you want to know how to make extra money, search for:
best adsense alternative Wrastain’s tools

Reply
asiel 6/04/2018 | 4:00

I need urgent help to make a triangular graphic with beta diversity components, I can upload my data matrix, if someone can explain how to do it.
my email asiel.cabrera@estudiantes.uo.edu.cu

Reply