Hi!
My study involves comparing patient and control samples, with each category having two kinds of RNA - one derived from blood serum and another from exosomes (from the same serum sample). Ultimately, I'm interested in comparing variations in gene expression not only between patient and control groups (in blood serum or exosomes separetely), but also between paired serum and exosomal RNAs samples within each group (paired patient samples and paired control samples). Given that exosomal RNA is part of the blood serum RNA, there's expected to be some overlap, but again it is different RNA isolation, so there can also be some variation. So my question is, can I normalize all four of these subsets (patient serum, patient exosomes, control serum, control exosomes) at once using DESeq2? Is it able to handle the variation in this case?

Design:

|name     |condition     |type    |paired |
|:--------|:-------------|:-------|:------|
|Sample01 |serum_patient |patient |S1     |
|Sample01 |sEVs_patient  |patient |S1     |
|Sample02 |serum_patient |patient |S2     |
|Sample02 |sEVs_patient  |patient |S2     |
|Sample03 |serum_patient |patient |S3     |
|Sample03 |sEVs_patient  |patient |S3     |
|Sample04 |serum_patient |patient |S4     |
|Sample04 |sEVs_patient  |patient |S4     |
|Sample05 |serum_patient |patient |S5     |
|Sample05 |sEVs_patient  |patient |S5     |
|Sample06 |serum_control |control |S6     |
|Sample06 |sEVs_control  |control |S6     |
|Sample07 |serum_control |control |S7     |
|Sample07 |sEVs_control  |control |S7     |
|Sample08 |serum_control |control |S8     |
|Sample08 |sEVs_control  |control |S8     |
|Sample09 |serum_control |control |S9     |
|Sample09 |sEVs_control  |control |S9     |
|Sample10 |serum_control |control |S10    |
|Sample10 |sEVs_control  |control |S10    |

To summarize, here are the comparisons:

1. serum_patient vs serum_control
2. sEVs_patient vs serum_control
3. serum_patient vs sEVs_patient
4. serum_control vs sEVs_control

Thank you very much!

You can use the controlGenes argument with estimateSizeFactors() to base normalization on a subset of genes that you believe are suited for this. Transcriptome, serum and exosome are likely of very different composition, so your knowledge of the biological system must be used to decide which genes you want to normalize to.