Seurat Analysis: Deleting cells from seurat object

Question

[odi]

I have not been able to code anything as I am in the thought process.. but I was wondering I have 3 samples: DP1, DP2 and DP3.

DP1 is tumor cells but both DP2 and DP3 are tumor + PBMCs. I would like to delete the PBMCs. How could one potentially go about this so that we have DP1, DP2 and DP3 as tumor cells only for further analysis? I have included a column in the metadata titles cells to differentiation DP1, DP2, and DP3.

I hope this makes sense. All help will be appreciated as I have no idea how to do this in seurat.

Answer 1

[jared.andrews07]

It depends what information you already have. If you can easily distinguish the PBMCs from tumor cells, e.g. by clustering and looking for known tumor markers, then you can just add a cell metadata column with a "tumor" or "PBMC" designation and subset based on that.

If you don't have a good way to designate a cell's origins, then you could try integrating all the samples together, which would likely force the PBMCs together and allow them to be easily removed. From experience, integration generally has a weaker effect on tumors than typical tissues due to greater inter-tumor heterogeneity, so I'd expect the tumor cells not to get crammed together quite as cleanly/closely.

[odi]

Thank you so much! I'll try this and see if any progress but this has been the most assistance i've had on this

[odi]

So, I have 3 samples:

DP1 = Tumor cells DP2 & DP3 = Tumor + PBMC cells

I took your suggestion and performed Seurat integration and created a dimplot split by cells (tumor cells, tumor+pbmc cells). I am attempting to remove the pbmc cells. Could i use the FindConservedMarkers to find the markers that are common to both dimplots?

I am assuming that this function would be able to only identify the tumor cells since that is what the dimplots would share in common... that way i can subset the seurat object based on these markers....

To be clear is that what the FindConservedMarkers function does or this thought process would be wrong? I'm attaching the umaps here.

[jared.andrews07]

See one of my other answers for a description of what FindConservedMarkers does and how it compares to the other marker functions in Seurat.