How many cells are needed in subcluster to draw conclusions?

Question

[S]

Hi,

I am new to immunology, so still learning a lot about the different cell types. I have a scRNA-seq dataset that I analysed using Seurat. Our cell type of interest were T cells, so I subsetted those after annotation.

Our main interest is in tregs, so my questions are:

1. We have ~4200 cells in the tregs subset (about 67k total T cells). That is ~7%. I felt like the numbers were slightly low (please correct me on this), and if they are not low then is it because they are a hard cell type to capture, or generally not as abundant?
2. I further subclustered tregs since I know there are different types. I found a subcluster of interest but it is only about 100 cells out of ~4200 (~2.3%). This is a really interesting cluster, but are 100 cells enough to draw any conclusions from? As in can I use this cluster at all for pathway analysis and stuff due to the really low cell count?

Thank you I appreciate the help!

[ATpoint]

This cannot really be answered. You are missing on the organ you assay, rpeparation method, maybe FACS enrichment scheme -- all of which influences cell frequency beyond the true frequency in that organ.

As for the subclustering questions, T cell research is no vacuum, others have published signatures and subsets of T cells before. Please read the literature and see whether you can match the identified cluster to anything that is known. Without knowing immunology and the literature well it is borderline impossible to judge whether what you see is a) already known b) potentially interesting, c) potentially meaningful d) in striking contrast to what is known and hence e) worth focusing on.

Also be aware that subclusters from RNA does not necessarily correlate to FACS-based clusters that are prominent in the literature.

[S]

Thank you for your reply! I don't have all the details of the experiment, the organ is gut (colon/ileum), unsure of the entire preparation method, but I know these are just immune cells - so post EDTA epithelial depletion.

And maybe I didn't phrase my question well, but I meant I know what T cell subset it is, and it's interesting since they are rare to capture. My question was whether you can statistically draw conclusions from subsets that have only 100 cells - is that enough power? Thanks again.