The example is split into 2 Parts:
Part 1 must be completed first to create a file,
SNPs_cleaned.csv, that has been completely prepared for
analysis.
Now in Part 2, you will analyze the data with PCA. The steps here will be:
scale())prcomp())screeplot())vegan::scores())In the code below all code is provided. Your tasks will be to do 4 things:
Load the vcfR package with library()
library(vcfR) # KEY
##
## ***** *** vcfR *** *****
## This is vcfR 1.13.0
## browseVignettes('vcfR') # Documentation
## citation('vcfR') # Citation
## ***** ***** ***** *****
library(vegan)
## Loading required package: permute
## Loading required package: lattice
## This is vegan 2.6-4
library(ggplot2)
library(ggpubr)
Set the working directory
Load the data
SNPs_cleaned <- read.csv(file = "SNPs_cleaned.csv")
list.files(pattern = ".csv")
## [1] "df.csv" "SNPs_cleaned.csv"
## [3] "walsh2017morphology.csv"
warning("If this didn't work, its may be because you didn't set your working directory.")
## Warning: If this didn't work, its may be because you didn't set your working
## directory.
TODO: scaling the data needs to happen in order to do PCA
SNPs_scaled <- scale(SNPs_cleaned)
TODO: use prcomp() to run PCA on the scaled data
pca_scaled <- prcomp(SNPs_scaled)
TODO: A screeplot of a PCA is used to determine the relevant PCs for further analysis
TODO: SNPs PCA
screeplot(pca_scaled,
ylab = "Relative importance",
main = "SNPS PCA")
TODO: The only big drop is from PC1 to PC2 which indicates that PC1 is the only one really worth interest to look at
TODO: using a function for variation explained based on 10 PCs. The horizontal line on the plot indicates the variation that would be explained if all PCs were equally important
summary_out_scaled <- summary(pca_scaled)
PCA_variation <- function(pca_summary, PCs = 2){
var_explained <- pca_summary$importance[2,1:PCs]*100
var_explained <- round(var_explained,1)
return(var_explained)
}
var_out <- PCA_variation(summary_out_scaled,PCs = 10)
var_out
## PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC10
## 19.9 2.3 2.2 2.1 2.0 2.0 2.0 1.9 1.8 1.8
N_columns <- ncol(SNPs_scaled)
barplot(var_out,
main = "Percent variation Scree plot",
ylab = "Percent variation explained")
abline(h = 1/N_columns*100, col = 2, lwd = 2)
TODO: PC1 explains over 19.9% of variation, which is much more then the other PCs
TODO: A biplot is made to show the relationships between the features
biplot(pca_scaled)
TODO: There are many dimensions to SNP data so biplots are not usually helpful because all of the ploted vectors create a mess to look at
TODO: get the PCA scores on the scaled data using vegan::scores()
pca_scores <- vegan::scores(pca_scaled)
TODO: create a vector with the label names
pop_id <- c("Nel","Nel","Nel","Nel","Nel","Nel","Nel","Nel",
"Nel", "Nel", "Nel", "Nel", "Nel", "Nel", "Nel", "Alt",
"Alt", "Alt", "Alt", "Alt", "Alt", "Alt", "Alt", "Alt",
"Alt", "Alt", "Alt", "Alt", "Alt", "Alt", "Sub", "Sub",
"Sub", "Sub", "Sub", "Sub", "Sub", "Sub", "Sub", "Sub",
"Sub", "Cau", "Cau", "Cau", "Cau", "Cau", "Cau", "Cau",
"Cau", "Cau", "Cau", "Cau", "Cau", "Div", "Div", "Div",
"Div", "Div", "Div", "Div", "Div", "Div", "Div", "Div",
"Div", "Div", "Div", "Div")
TODO: create a dataframe by combining the PCA scores and the population labels
pca_scores2 <- data.frame(pop_id,
pca_scores)
TODO: plot of the PCA scores colored according to population
TODO: PCA SNP Scatterplot TODO: UPDATE X and Y AXES WITH AMOUNT OF VARIATION EXPLAINED
ggpubr::ggscatter(data = pca_scores2,
y = "PC2",
x = "PC1",
color = "pop_id",
shape = "pop_id",
xlab = "PC1 (19.9% variation)",
ylab = "PC2 (2.3% variation)",
main = "PCA SNP scatterplot")
TODO: We can see two distinct groups along the PC1 axis, however there is no clear groupings along the PC2 axis