Hi all,
I feel totally ill-equip to ask this question that I know is likely a "bad question", so I am hoping for an educational moment here.
I have a set of spatial transcriptomic data that we are trying to squeeze like a dirty dish rag for any salient information. These data are one expt and one ctrl mouse brain section on the 10x platform. They are loaded as a seurat object. They were subset to the ROI, normalized using SCTransform V2 and integrated according the the respective seurat vignettes.
One such "ringing of the dish towel" is looking for genes that co-express with known neuronal activity markers (immediate early genes). My approach to this was rudimentary and simplistic (as was my bioinformatics training). I extracted the counts information from each condition, computed a whole pairwise correlation matrix, and subset this square matrix to see which genes had a correlation coefficient near 1 for each of the activity marker genes. My rationale was "these are the genes that are positively correlated in their expression with the activity markers. Perhaps, the genes positively correlated genes in the expt condition are different than those in the ctrl condition."
My first ELI5 question is, is this even a valid approach? If not, what would be a more valid approach for this line of reasoning? Obviously the data are beyond weak with an n of 1, DE analysis is largely ineffective and WGCNA is not compatible with a data set of this size.
My further line of questioning (and what brought me here in the first place) is: is there any way to normalize my data relative to one of these activity marker genes so the correlation coefficients are more comparable between conditions? For example, could I use the var.to.regress method in seurat's SCTransform to control for the expression of one of these activity markers, so that the the levels of coexpression between conditions are starting from the same baseline?
I look forward to being torn apart academically, and hope to come out on the other side with some answer and enlightenment. Thank you in advance
