[R] Plot Principal component analysis

[Thibaut Jombart]

Jim Lemon wrote:

>SNN wrote:
>  
>
>>Hi,
>>
>>I have matrix of 300,000*115 (snps*individual). I ran the PCA on the
>>covariance matrix which has a dimention oof 115*115. I have the first 100
>>individuals from group A and the rest of 15 individuals from group B. I need
>>to plot the data in two and 3 dimentions with respect to PC1 and PC2 and (in
>>3D with respect to PC1, PC2 and PC3). I do not know how to have the plot
>>ploting the first 100 points corresponding to group A in red (for example)
>>and the rest of the 15 points in Blue? i.e I want the each group in a
>>diffrent color in the same plot. I appreciate if someone can help.
>>
>>    
>>
>Hi Nancy,
>(if indeed you are a Nancy and that is not a webnym)
>Say that your groups really are coded "A" and "B", the group coding 
>variable is called "group". You can define a color vector like this:
>
>colorvector<-ifelse(group=="A","red","blue")
>
>Jim
>
>______________________________________________
>R-help at r-project.org mailing list
>https://stat.ethz.ch/mailman/listinfo/r-help
>PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>and provide commented, minimal, self-contained, reproducible code.
>
>
>  
>
Hi Nancy,
in your case you may also use inertia ellipses to represent your groups, 
in addition to different colors.
Here is an example using a microsatellite dataset from adegenet (but 
valid for SNPs of course):
#####
library(ade4)
library(adegenet)
data(microbov) # dataset

# replace missing values
obj=na.replace(microbov,method="mean")

# perform your pca, keep 3 axes
pca1=dudi.pca(obj$tab,scannf=FALSE,nf=3,scale=FALSE)

# plot the result
s.class(pca1$li,obj$pop)
s.class(pca1$li,obj$pop,col=sample(colors(),15)) # here, replace "col" 
by the appropriate vector of colors.

#####
The resulting graphic represents each genotype by a point, and adds 
ellipses of different color for each group; each ellipse represents 95 % 
of the inertia of the corresponding group. The more ellipses overlap, 
the less your groups are differentiated on the factorial plane.

Cheers,

Thibaut.
-- 
######################################
Thibaut JOMBART
CNRS UMR 5558 - Laboratoire de Biométrie et Biologie Evolutive
Universite Lyon 1
43 bd du 11 novembre 1918
69622 Villeurbanne Cedex
Tél. : 04.72.43.29.35
Fax : 04.72.43.13.88
jombart at biomserv.univ-lyon1.fr
http://lbbe.univ-lyon1.fr/-Jombart-Thibaut-.html?lang=en
http://pbil.univ-lyon1.fr/software/adegenet/