I have RNA-seq data from a cell line with a knockout of a gene involved in miRNA processing. We suspect that this mutation causes global downregulation of most genes. If this is true, the DESeq2 assumption used for calculating size factors (that most genes are not differentially expressed) would not be satisfied.
Additionally, we suspect that even "housekeeping" genes might be changing.
Unfortunately, repeating the RNA-seq with spike-ins is not feasible for us. My question is: Could we instead use a spike-in normalization approach with the existing samples by measuring the relative expression of selected genes (e.g., GAPDH) using RT-qPCR in the parental vs. mutant cell line, and then adjust the DESeq2 size factors so that these genes reflect the fold changes measured by qPCR?
I've found only this paper describing a similar approach. However, the fact that all citations are self-citations makes me hesitant to rely on it.
The question is what the point is here. If you have global downregulation then everything is lower. Do you really need per-gene differential analysis to show this? Arguably, you could demonstrate this much simpler with a well-designed qPCR panel which, unlike RNA-seq can give you absolute quantification. And if you're more interested in showing relatively which genes are affected most, then you can still do a standard DEG analysis and enrich for genes that have large fold changes and a lot of evidence for differential expression, ignoring the global levels.