Filtering genes with limma-voom

Question

[im]

I am trying to follow this tutorial on TCGA-LUAD RNA-Seq data, but I am having trouble coming up with an appropriate cutoff to produce a "good" voom plot. I was doing some research on how they should be chosen, and since the filtering is generally done using cpm, the cutoff depends on library size. However, I am having some trouble finding how it should relate to library size - does anyone mind helping me out with what I should try?

[h.mon]

edgeR has a function to perform automatic filtering, see ?filterByExpr. I guess filterByExpr is not used in that tutorial as it was (probably) introduced after the tutorial was published. There are several posts / tutorials incorporating filterByExpr() in their workflow, for example: RNA-seq analysis is easy as 1-2-3 with limma, Glimma and edgeR.