PCA of variable features vs all genes in integrated analysis

Question

[aa123]

In this vignette (ctrl vs stim), does PCA use the 2000 variable features or use all the genes from the integrated analysis?

https://satijalab.org/seurat/v3.2/immune_alignment.html

Thank you.

Answer 1

[ATpoint]

It is the variable features identified in this command:

ifnb.list <- lapply(X = ifnb.list, FUN = function(x) {
    x <- NormalizeData(x)
    x <- FindVariableFeatures(x, selection.method = "vst", nfeatures = 2000)
})

The idea is to first identify genes with large variance assuming that these will drive separation between cells, then reduce them into principle components and then perform integration in PCA space.

[aa123]

Thank you.

Is this because the default assay is “integrated”?

The RunPCA command does not specify using variable features through features=VariableFeatures(object=)).