Hello, I am generating DEGs using results() and lfcshrink(). I would like to use logFC from lfcshrink() for MA plot and heatmap of top 30 DEGs with the highest logFC DEGs (since the user manual suggested that shrinking logFC is useful for visualisation and ranking).

And then I use DEGs from results() with p<0.05 and abs(logFC)<1 for further enrichment & pathway analysis (since the low expressed genes might be relevant too, but I do not know, therefore don't shrink the logFC to not lose DEGs)

I wonder if this is a good approach or should I use lfcshrink() for all downstream analysis too?

Your approach sounds fine to me.

You can do:

res <- results(dds, ...)
lfc <- lfcShrink(dds, ...)
res$posteriorLFC <- lfc$log2FoldChange

And then all the information is tied to res.