I have 3 fish groups differently treated. With scRNA-seq and data analysis using Seurat, I found out there a big variation in the cell numbers/treatment. Given that the batch effect was controlled, and this variation is demonstrating the same pattern across all the replicates, what sort of analysis in Seurat can support or refute that this finding is attributed to the different treatments?
Do you mean the actual cell numbers found in each sample or the cell type proportions observed in each? For the former, you are much better off using experimental methods to confirm this (e.g. flow cytometry, luciferase imaging, etc if cells are in vivo treated or growth curve assays if in vitro treated), as you have no way to confirm this from the single cell experiment itself (it could have been technical variation, different numbers of cells loaded by accident, etc).
For the latter, I'm going to refer to the differential abundance chapter from the OSCA book, which does exactly that via edgeR. While that book deals with SingleCellExperiment objects, you should be able to extrapolate its examples to your Seurat object fairly easily since you're just dealing with metadata. These types of analyses can still be subject to sampling biases, so you should ensure you have several replicates for each condition. And again, confirming with experimental methods is always ideal.
Hi Jared,
I mean the actual cell numbers found in each sample, and the cells are in vivo treated. That's right, a technical variation or other accidental error might have happened. Do you not think that the number of reads and the reads vs UMI histogram could help in figuring the driver of this variation out? If yes, do you have an idea how to interpret this information?
I appreciate your answer too much, many thanks
Log in to answer this question.
