Seurat subset cluster cell annotation

Question

[leranwangcs]

Hi,

I used Seurat to do clustering, then singleR to do cell annotation. In this annotation, I got multiple cell types such as B cells and T cells and Epithelial cells. I'm only interested in Epithelial cell, so I subset the Seurat object to only keep Epithelial cell. Then I redo clustering on Epithelial cell cluster. Here is the new clusters Seurat generated:

I want to assign cell names to each cluster, but I'm not sure how to do this on the further clustered data (I don't think I can use singleR again, because if SingleR can annotate these clusters, it should have done that at the first time I used it.)

I used seurat to calculate the gene expression in each cluster and hoping to get cell type based on the gene expression information. The gene expression table looks like this:

So my question is, does anybody know how to do cell annotation on these sub clusters? Is there any reference database that I can use to find cell names based on the gene expressions?

Thanks so much! Leran

[Pratik]

Granted you played around/optimized clustering in your first round prior to SingleR, and also granted you have exhausted the currently available pre-loaded annotations in celldex, and also making sure you took into account your tissue and, if applicable, disease state, time point, etc...

You could maybe load in your own reference dataset into SingleR if you find someone else has already done a similar tissue/disease/time/etc in another study, and use that to annotate your dataset. (I have not done this myself yet, but I believe it's pretty doable)

Lastly you could do it the old-fashioned way and name clusters based on markers... Might be long reads but these posts could help: General question about clustering in scRNAseq and how to find each cluster in single-cell represent which cell type?