How should I normalize my RNAseq data? I appreciate if someone feels they can help me with this. I will try to describe my experiment and hopefully the information will be a little clearer. I have run a polysome profiling followed by deep RNA sequencing. During this experiment, a cell lysate is divided according to the number of ribosomes. I have 3 different conditions, including control, untreated mutant and treated mutant. In addition, a three-level factor (monosome, light polysome and heavy polysome), which means different fractions associated with a different number of ribosomes, we have run it in triplicate, so we have 27 samples in total. We have added spike In RNA in the samples at a known concentration in relation to the total amount of RNA. This particular experiment involves some peculiarities, so I have doubts about what could be the best method to normalize the data: * samples from monosomes, light polysomes and heavy polysomes, are associated with a different number of ribosomes so the mRNA/rRNA ratio can be significantly different. * The RNA composition is expected to be very different between monosomes, light polysomes and heavy polysomes, even if the fractions come from the same sample * We could have a lot zero value for many genes according to their mRNAs distribution in the polysome fractions. I understand that DEQSeq2 generates a pseudo sample by geometric mean calculation, could it be a issue to use this algoritm? * We want to perform a differential expression analysis, comparing the levels of expression in monosome, light polysome and heavy polysome, treated and untreated mutant group vs control group. But I would also like to make a differential expression analysis including the levels of the same sample for example using monosomes as a reference value. Do you think that in this case it is a good idea to normalize using Spike-In RNA? Can you suggest me a DESeq2 formule?

I think this would be my advice. Note that there is an assay specific term in the model:

https://support.bioconductor.org/p/61509/