How to highlight specific cells in Seurat 2.4

Question

I used Seurat 2.4 on our scRNA dataset to obtain the following tSNE plot.I was able to successfully extract cell IDs from the different clusters, and generate gene expression profiles. The analysis, and the biology makes sense. I have been however stuck in trying to highlight specific cells we are interested in using the Cell IDs (barcodes). For e.g, in this plot I am hoping to highlight specific cells (~ 120 of them) within Cluster 5 (blue cluster on the top). It can be either in featureplot mode or in this plot itself by an overlay, it doesn't matter. All I have to show are the 120 cells within the cluster. For eg. if cluster 5 has cells atag-1 atgc-2 atat-3, cacc-4 cat-5... i want to recreate this plot using atag-1 atgc-2, cacc-4

I tried using do.identify = TRUE but that is a visual selection and can select regions, but is not useful to select specific cells. I was thinking if there is a way to subset and replace cluster 5 to include just the cells (~120 odd out of the total 200) but was unable to crack it. I tried by using cells.use but then realized that doesn't work either.Any ideas will be greatly appreciated, to get me out of this rut. Thank you

Answer 1

You can "highlight" any cells you'd like by creating a new column (or modifying an existing column) in the meta data slot within the Seurat object. From there you can modify the contents of that vector to your liking and plot the results.

It's not clear if you've been successful in obtaining the barcodes for the subpopulation to begin with, because that's a separate issue all together, and you need to clarify what attribute you are trying to subset the cells on because there are a number of ways to accomplish that. If you've already obtained the barcodes then I'll describe how you would plot the results below.

Let's assume you'd like to keep the original cluster IDs for all other cells. We'll start off by duplicating the cluster ID column (this isn't necessary, but it's good practice to preserve the original IDs).

n <- [email protected] # for convenience 
n$new_column <- n$res.1.0

Now you need to modify the ids for cluster 5 to break them up into 2 populations.

n[rownames(n) %in% my_barcodes, "new_column"] <- "5b"

You can plot the new column using DimPlot with the argument group.by = "new_column", which will make the subpopulation stand out from the rest of the cells in cluster 5.