Gene list for GO enrichment after DESeq2. Should I use a gene list of differentially expressed genes (DEG) before or after shrinkage?

Question

[Lautaro]

Hi, I have been starting by first steps into RNA-Seq data and I have been performing my first analysis since two months ago. So I apologise if the question is kind of weird. I have run the DESeq2 analysis as it is detailed in the vignette and the bioconductors webpages. So far, I understand that the genes I get after the shrinkage are a more conservative list of genes that the ones before the procedure. On top of that, after I shrink the data I filter the output with the following line to keep those genes within a particular range of FC.

resLFC_Ordered <- resLFC[order(resLFC$pvalue),]
resLFC_Ordered<- subset(resLFC_Ordered, padj <= 0.05 & (log2FoldChange <= 10 & log2FoldChange > 1.5 | log2FoldChange >= -10 & log2FoldChange <= -1.5))

My question is: is it OK to input into enrichGO this list of genes? Or should I use the list of DEG before the shrinkage? When moving forward into the analysis and deciding which genes are we going to move into the wetlab, we will be focusing in the list after shrinkage so my guess is to input it also into enrichGO.

Thanks!

Answer 1

[Carlo Yague]

My advice would be to use the list with the shrinkage, but it should not change your results drastically. Indeed, shrinkage has a strong effect only on lowly expressed genes.Differential expression is likely not significant for such genes, so they get filtered out with the p.adj<0.05 threshold anyway.

[Lautaro]

Thanks for the answer!