No GO term enrichment in differentially expressed genes

Question

[gopal2706]

I am analyzing an RNA-seq dataset (with biological triplicates) of a transfection experment. Using DeSeq2, i obtain a list of genes which are differentially expressed across the samples (vs empty vector in each case). However when i try to do a GO-term analysis, i obtain no enrichment. What is more stranger is that in multiple orthogonal experiments i obtain a very similar gene list. This gene list is also populated by a lot of noncoding RNA, pseudogenes etc (all above a +/-0.75 log2 cutoff). I am wondering if this is a general transcriptional response to Lipofectamine or am i looking at garbage? Any help would be appreciated

[Jean-Karim Heriche]

You don't give much details so we can't give you specific advice.
If your controls are treated the same way as your samples, the differentially-expressed genes should reflect the samples treatments (this is what controls are for). Cells definitely respond to transfection, this is why the controls must also be transfected with empty vector/neutral reagent (you did do this, right ?). So if you compare similarly transfected control and sample, we would not attribute any differential effect to the transfection.
If different types of experiments confirm the same gene list, I would tend to say this is a robust result.
There are plenty of good reasons why you may not find enriched GO terms. One is that there is simply no enrichment (this can happen in particular if your genes are not well annotated especially if you include non-coding junk genes), another is that the analysis method you're using may not be sensitive enough.
Finally keep in mind that statistical significance bears no relation to biological relevance.