Grouping datasets in Seurat

Question

[jordi.diaz-manera]

Hi. I am working with scRNAseq data from 5 different samples (3 patients and 2 controls). I am able to integrate them, create a Seurat file and obtain nice clusters. However I am struggling to group the 3 patients and the 2 controls together to compare gene expression between patients and controls. For example if use VlnPlot to show expression of one and I write the command group.by="orig.ident" it shown the expression of the gene by each 5 condition (patient 1, patient 2, patient 3, control1 and control 2). I am unable to find out how to show gene expression by patient vs control. Any idea?

Answer 1

[fracarb8]

You need to add a column in the metadata ([e-mail]) containing your grouping, so that each UMI are classified. For example:

control_samples <- c("C1","C2")
[e-mail]$condition <- sapply([e-mail]$orig.ident, function(ita) ifelse(ita %in% control_samples,"Control","Patient")

Then you can group.by="condition"

[jordi.diaz-manera]

Thanks a lot for your answer. That worked perfectly well!!

Thanks again

[jordi.diaz-manera]

That worked perfectly well!. Thanks a lot

[rpolicastro]

Just a general comment but seuratObject$condition works as well since they defined Seurat specific methods for $ and $<- to set and get metadata. Generally you don't want to access slots in S4 objects directly since by R convention the slots can change without documentation, but the generics and methods setting and getting slots should in theory always point to the correct location.

[fracarb8]

I do agree with you, but since the painful migration from seurat 2 to seurat 3 I prefer to point directly to the slots to retrieve data. Between versions, the seurat structure has not changed that much, while the specific methods have.