Deseq2 design clarification

[rammohanshukla]

Hi, I am working with human postmortem samples wherein I have 5 groups of subjects:

A. Control: 20 samples
B. First episode: 20 samples
C. First episode Remission: 15 samples
D. Second episode: 20 samples
E. Second episode Remission: 15 samples


Note that the progression of A to B to C to D to E is expected to be in phase (something like sinusoidal curve)

I also have other parameters for the samples, which include: Age, Sex, RIN value, PMI (postmortem interval), pH of the sample


I wish to test the difference of the 5 groups in ANOVA like analysis in DEseq2 but after controlling for Age, Sex, RIN value, PMI (postmortem interval), pH of the sample

My phenotype table looks like this:

Sample Age Sex Rin PMI pH DiseaseStage
1 binned factor binned binned binned Control
2 Control
3 First episode
. First episode
. First episode
. .
. .
90 Second episode Remission


This is my design
dds <- DESeqDataSet(se, design = ~ Age+Sex+RIN+PMI+pH+DiseaseStage)

dds <- DESeq(dds, test = "LRT", reduced = ~ Age+Sex+RIN+PMI+pH)
results(dds)

I am hoping that I get 1 p-value for ANOVA like comparisons between the disease stage after controlling for Age, Sex, RIN, PMI and pH
Also, I would like to know how many variables I can control for in the design? For instance, here I am using 5 (Age, Sex, RIN, PMI and pH) is it okay? If yes, is there any rule to how many variables to control for in the design?