Hi all, A question came to my mind when analysis gene expression using RNA-seq. After using DEseq2, a large number of genes were listed as differential expressed (e.g. FDR<0.01 or 0.05, absolute of log2 fold change >1). When took look at the FPKM of those genes, lots of them were very small. I think took all of them for further analysis is time consuming and also included a bunch of noises.
I saw a paper use the FPKM of one housekeeping gene they used to do QPCR as the cutoff. (http://aem.asm.org/content/79/7/2397.abstract)
In my opinion, use the mean FPKM (total FPKM of genes/gene number) of each condition as a cutoff/threshold might be a choice.
Any other thoughts on this issue are greatly appreciated!