I was wondering if there is a possibility to skip the clustering step in Seurat and just do the differential gene expression as I know what are the cell populations I've used as my input data. I don't want Seurat to find any novel subpopulations of the cells but I just want it to find the differential gene expression between 2 different populations of the cells in my input data.
I think you can perform differential expression analysis without performing the initial PCA, but you have group cells based on some gene expression pattern.
Sure, if you can assign whatever labels necessary to each cell (by barcode) to group them as necessary in the metadata of the Seurat object. Then you can just run FindMarkers between the two groups after setting the Ident of each cell to that metadata variable.
So in short, the commands you want are AddMetaData, StashIdent (to save the original barcodes to a new metadata column for posterity), Idents (to set the cell identities to whatever group variable you added with AddMetaData), and FindMarkers (to actually perform the differential expression between the groups).
my input is a read count matrix into Seurat. I'm making this into a Seurat object initially, so to this I do AddMetaData to say which cell type this is, then add Idents and use them in Findmarkers. Is this correct?
I think you can perform differential expression analysis without performing the initial PCA, but you have group cells based on some gene expression pattern.