Best package to cluster a single cell dataset using marker gene lists retrieved from elsewhere?

Question

[A248]

Say for example we have a PBMC single cell dataset generated in the lab, and would wish to use differentially expressed gene lists of different cell types (B, T, monocyte etc) which have been published elsewhere to guide the clustering of our single cell dataset.

Is there a package which would be best suited for this purpose?

Thanks

[paria]

Hi @jared.andrews07. To pick up on this question shall I provide the reference database for using SingleR? As I am trying to annotate cell type in my datasets which I don't have a reference for it. I appreciate any help. Paria

[GenoMax]

Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized. This comment should go under @Jared's answer.

SUBMIT ANSWER is for new answers to original question

[jared.andrews07]

I recommend opening your own question, but in short, SingleR requires reference datasets. If you don't have one for your cell types of interest, you'll have to figure out some other way to annotate your cells (some other options shown here).

Answer 1

[jared.andrews07]

I'm going to plug SingleR for this (full disclosure: I've been involved in its development). It will annotate your cells based on purified bulk or single-cell reference datasets, and it has several immune datasets built in that should easily handle PBMCs. It's also very easy to use and is work-flow agnostic - you can use it with pretty much any other packages (Seurat, scater, SingleCellExperiment, etc). As long as you have a count matrix, you can use it.

It's quick, it's easy, it has diagnostic visuals to ensure the results make sense, and it doesn't rely on users coming up with dependable marker sets themselves.

[DeadlyFall]

I'll add my recommendation for this package. I've used it on several different projects using Seurat objects. Very simple to use and does a great job, especially with immune cells. Thank you for helping to make this package!