Hi, I'm having trouble figuring out the appropriate design for my data matrix. My data look something like this:

Condition   Sex     MouseID NewMouseID
WT          Female  1       1
WT          Female  2       2
WT          Female  3       3
WT          Female  3       3
WT          Male    4       1
WT          Male    5       2
WT          Male    6       3
WT          Male    6       3
KO          Female  7       1
KO          Female  8       2
KO          Female  9       3
KO          Female  9       3
KO          Male    10      1
KO          Male    11      2
KO          Male    12      3
KO          Male    12      3

My problem is with multiple samples coming from the same mice (note MouseIDs 3, 6, 9, and 12). If possible, I would like to not simply sum/average those samples and instead include them as a nested factor. The reason for this is those duplicates were from different sites within the same mouse. I had previously averaged technical replicates (same site, sequenced 2+ times).

If I do:

design = ~ MouseID + Sex + Condition

I get an error. My understanding is that this doesn't work because it treats the MouseIDs at 12 different "batches" and it would be impossible to tell if differences were due to the "batch" vs. the Condition.

Instead, my thought is to include a "NewMouseID" factor, that sequentially numbers the mice within each "group (i.e. each Condition/Sex combination). Is it possible to include this information in the design matrix? I want to control for differences between samples within the same mouse, as well as differences between male/female. However, I do want to note that there is no specific relation between MouseID #1 and MouseID #4, for example, despite them having the same NewMouseID.

Thanks in advance.

You don't need to control for sex because controlling for mouse is more fine-grained. When you give each mouse a baseline, this takes care of any differences across sex because the mice are nested within sex.