identification of rare cell population in Sc-RNA sequencing

Question

[Maria17]

Greetings,

I have single-cell sequencing data from three different Patients(hepatocellular cancer). Are there any new analysis techniques specific to identify the Rare cell population? I've already performed quality control and clustering steps using Seurat and Scater and Scran, and I'm not sure how to proceed or do the next steps.

publications, machine learning algorithms, or other statistical analyses (especially R packages using Seurat or SCE object) would be appreciated.

Thanks

[ATpoint]

What do you mean by the rare cell populations? Did the clustering not produce what you expected?

[Maria17]

not exactly. I need to identify the Rare cell population and Assigning cell type identity to clusters but before that, I need a strong clustering method. the current methods that I used were not suitable to identify the rare populations.

[ATpoint]

Maybe there are none? An evidence that The Rate cell populations (still do not exactly know what you mean), are present? The established clustering methods are quite "good" but no method will work magic. Please give details.

[Maria17]

clustering from UMAP aims to identify the major populations. There may be rare cell populations that cannot be identified with this UMAP/clustering approach. We don’t know if such cells exist, nor do we know what they are. One possible example is the so-called cancer stem cells, which may be a quiescent cancer cell population (which is typically highly proliferative). Another possible example is very specific/specialized immune cell populations that may be a subset of a main immune cell population (e.g. T-cells). These are only two possible examples, we do not know if there are rare cells, nor do we know what their identities could be.