200 accessions to achieve reasonable statistical power. The information on each accession can be found in the 1001 Genomes
website: https://1001genomes.org/.
3.3 Evaluation of
Population Structure
Principal component analysis (PCA) can be used to evaluate
subpopulation-specific variations in allele distribution of the
SNPs, which results in “population structure” [20]. If population
structure exists among accessions, they show clusters in the PCA
plot. In this protocol, we will use EIGENSTRAT [21] implemented in PLINK and R to visualize the PCA plot.
3.3.1 Calculate
Eigenvectors and
Eigenvalues with PLINK
EIGENSTRAT is a PCA-based method used to evaluate population
structures in GWAS populations. Since PLINK version 1.9,
EIGENSTRAT has been implemented and can be easily used to
draw PCA plots of accessions.
$plink --file 1001genomes_snps_maf0.1_ACGTN --pca 20 header tabs var-wts
output files: plink.eigenval, plink.eigenvec, plink.eigenvec.var
3.3.2 Draw PCA Plot
1. Run R to read in the eigenvalue and eigenvector file.
>eigenvec_table <- read.table("plink.eigenvec",header=T) #read eigenvec file
>egval <- scan("plink.eigenval") #read eigenval file
2. Draw a PCA plot.
> png(filename="accessions_pca.png")
> par(mar=c(6.1,6.1,2.1,2.1)) #give margins to the figure
>plot(eigenvec_table[3:4],pch=20,xlab=paste("eigenvector1\n",egval[1], "% of observed
genetic variation", sep=""), ylab=paste("eigenvector2\n",egval[2], "% of observed genetic
variation", sep=""))
>dev.off()
Then, an image file named “accessions_pca.png” will be available. However the output image file can be in other formats such as
.svg or .pdf for various usage by exchanging the command line
png(filename="accessions_pca.png") to the following:
>svg("accessions_pca.svg",width=12,height=12)#out in .svg
>pdf("accessions_pca.pdf",width=12,height=12)#out in .pdf
The genetic distances of the accessions are plotted in a
two-dimensional space. If a sampled population does not have
strong population structure, it will be dispersed in the space,
194
Tak Lee and Insuk Lee
website: https://1001genomes.org/.
3.3 Evaluation of
Population Structure
Principal component analysis (PCA) can be used to evaluate
subpopulation-specific variations in allele distribution of the
SNPs, which results in “population structure” [20]. If population
structure exists among accessions, they show clusters in the PCA
plot. In this protocol, we will use EIGENSTRAT [21] implemented in PLINK and R to visualize the PCA plot.
3.3.1 Calculate
Eigenvectors and
Eigenvalues with PLINK
EIGENSTRAT is a PCA-based method used to evaluate population
structures in GWAS populations. Since PLINK version 1.9,
EIGENSTRAT has been implemented and can be easily used to
draw PCA plots of accessions.
$plink --file 1001genomes_snps_maf0.1_ACGTN --pca 20 header tabs var-wts
output files: plink.eigenval, plink.eigenvec, plink.eigenvec.var
3.3.2 Draw PCA Plot
1. Run R to read in the eigenvalue and eigenvector file.
>eigenvec_table <- read.table("plink.eigenvec",header=T) #read eigenvec file
>egval <- scan("plink.eigenval") #read eigenval file
2. Draw a PCA plot.
> png(filename="accessions_pca.png")
> par(mar=c(6.1,6.1,2.1,2.1)) #give margins to the figure
>plot(eigenvec_table[3:4],pch=20,xlab=paste("eigenvector1\n",egval[1], "% of observed
genetic variation", sep=""), ylab=paste("eigenvector2\n",egval[2], "% of observed genetic
variation", sep=""))
>dev.off()
Then, an image file named “accessions_pca.png” will be available. However the output image file can be in other formats such as
.svg or .pdf for various usage by exchanging the command line
png(filename="accessions_pca.png") to the following:
>svg("accessions_pca.svg",width=12,height=12)#out in .svg
>pdf("accessions_pca.pdf",width=12,height=12)#out in .pdf
The genetic distances of the accessions are plotted in a
two-dimensional space. If a sampled population does not have
strong population structure, it will be dispersed in the space,
194
Tak Lee and Insuk Lee