SCTransform in single cell

Question

[Omics data mining]

I have 4 samples (2 replicate each) sequenced with scRNAseq technology (10x Genomics) .

I am following these steps in seurat:

-> create an object from each of them separately with CreateSeuratObject,

-> merge the 4 objects with "merge" function.

-> perform SCTransform on the resulted object containing the group1, group2 ,group3 and group4

All samples falls into :

condition <- c("Tr", "Tr", "Ct", "Ct")

IDs <- c("a1", "a2, "b1", "b2")

Would it be correct if I include "condition" variable during the normalization ?

 SCTransform(object = object , vars.to.regress = c("IDs", "condition" )

Please correct me if I am doing something wrong. I am really confused for the normalization step.

I will appreciate all the suggestion.

Answer 1

[Rafael Soler]

Hello,

The option vars.to.regress in the SCTransform formula is used to regress out the effects of any covariates that have an effect on our data. In your case, IDs is impossible to regress out because they are the IDs of the samples. In the case of condition, probably you are not interested in eliminating the covariate either, since you will be interested in seeing what differences there are in the cell population of the treatment vs. control condition. In the vars.to.regress option, the effect of the mitochondrial or ribosomal genes could be put, if we saw a clear effect of them in the distribution of the PCAs.

Best,

Rafa