I am working with a set of samples and around 4000 genes being expressed across these. (no control).

• I have a counts df with samples as columns and genes as rows
• I have a coldata df = samples as rows and observations as columns

I have plotted a PCA using the DESEq

plotPCA(vsd, intgroup=c("condition", "type"))

pcaData <- plotPCA(vsd, intgroup=c("condition", "type"), returnData=TRUE)
percentVar <- round(100 * attr(pcaData, "percentVar"))
ggplot(pcaData, aes(PC1, PC2, color=condition, shape=type)) +
  geom_point(size=3) +
  xlab(paste0("PC1: ",percentVar[1],"% variance")) +
  ylab(paste0("PC2: ",percentVar[2],"% variance")) + 
  coord_fixed()

I would like to plot a biplot on the PCA using my coldata environmental variables so they point explaining the variability.

• Is there a way I could include the eigenvectors or the factors from my "coldata" which are potentially explaining the PCA variability of my samples?
• In what format do I need to have my data to (ie. only numeric, transpose, merge coldata+count data?

library(PCAtools)
biplot(pcaData)
# Error in nrow(y) : argument "y" is missing, with no default
sessionInfo( )
R version 4.0.3 (2020-10-10)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Catalina 10.15.4

Hi, PCAtools developer here. You first need to perform PCA via the PCAtools::pca() function. Please refer to my extensive vignette: https://bioconductor.org/packages/release/bioc/vignettes/PCAtools/inst/doc/PCAtools.html

More specifically, you will probably need:

p <- pca(assay(vsd))
biplot(p)

Note that DESeq2 also provides a PCA function.

Kevin